This systematic review and meta-analysis followed PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analysis) guidelines, as shown in Multimedia Appendix 1 [25], and was registered with PROSPERO (Prospective Register of Systematic Reviews; CRD42023443926).

The search strategy was developed based on a thorough review of existing literature and the collective expertise of our research team. The team members had undertaken specialized coursework in systematic reviews and meta-analyses and participated in multiple meta-analysis projects. Under the guidance of the principal investigators (LY, CEG, and EG), we performed exhaustive search in PubMed, Embase, Cochrane Library, and Web of Science, covering publications from the start of each database until October 2023.

Our search methodology combined specific keywords and Medical Subject Headings (MeSH), such as “sedentary behavior,” “mHealth interventions,” and “older adults,” as detailed in Multimedia Appendix 2.

Eligible studies met the criteria listed in Textbox 1.

All studies identified were uploaded to the Endnote X9 library (Clarivate) and underwent deduplication. Next, independent reviewers (SC and YY) independently screened the identified studies’ titles, abstracts, and full texts for potential eligibility based on predefined inclusion and exclusion criteria. After discussion and consensus, irrelevant articles were excluded. Data were then extracted using a standardized Microsoft Excel (version 16.78.3) spreadsheet. We extracted essential details such as author, year, country, participant ages, participant gender, disease, study type, sample size, intervention setting, intervention duration, intervention device, sedentary behavior assessment tools, primary assessment metrics, theories, and BCTs used in the study. Two reviewers performed data extraction independently and cross-checked each selected study to validate accuracy. Discrepancies in the screening process were resolved through consensus or consultation with a third reviewer (LY).

The methodological quality of the included studies was assessed using the Cochrane Risk of Bias (ROB 2) tool for RCTs and the Methodological Index for Non-Randomized Studies (MINORS) tool for nonrandomized studies. For RCTs, 5 domains were evaluated using ROB 2 by the Cochrane Handbook [28]: randomization process, deviations from intended interventions, missing outcome data, measurement of the outcome, and selection of the reported result. Each domain was rated as low risk, some concerns, or high risk of bias. For nonrandomized studies, MINORS was used to assess 12 items (eg, clearly stated aim, prospective data collection, and unbiased end-point assessment), each given a score of 0 (not reported), 1 (reported but inadequate), or 2 (reported and adequate) [29]. For noncomparative studies, 8 items are assessed, with a maximum score of 16. Scores between 13 and 16 indicate high quality, 9-12 moderate quality, and 5-8 low quality [29,30]. Two researchers (SC and AK) independently conducted the assessments, with cross-checking results. Disagreements were resolved through discussion or consultation with a third reviewer (LY).

Summary estimates were calculated for primary outcome variables measuring sedentary behavior, such as minutes spent sitting per day. The inverse variance method was applied using both random-effects and fixed-effects models based on the level of heterogeneity among studies [31]. When outcome measures across studies varied in terms of units or methods, standardized mean differences with 95% CIs were calculated based on mean values and SDs. Conversely, when studies reported outcomes using consistent measurement units and methods, mean differences were used to compare the intervention effect directly. When mean values and SDs were not reported, the missing data were obtained by contacting the original authors.

Meta-analyses were performed using Review Manager v5.4 (Cochrane Collaboration) [28], and the I² statistic was used to assess heterogeneity across the studies. A fixed-effect model was used when heterogeneity was not significant (I²50%), while a random-effects model was used if significant heterogeneity was present (I²>50%).

The PRISMA flowchart in Figure 1 details the process of identifying eligible studies. The initial database search identified 1870 articles from Web of Science (n=118), Embase (n=380), PubMed (n=590), and Cochrane (n=782). After removing duplicates, 1528 articles were excluded. Further screening of titles and abstracts removed 287 articles that did not meet the inclusion criteria. A full-text review led to an additional 42 articles being excluded due to their failure to meet the necessary criteria, leaving 13 articles for analysis. Three articles explicitly related to eHealth were excluded [32-34], resulting in 10 articles for systematic review [35-44]; 7 of 10 articles were excluded from the meta-analysis. The reasons for exclusion were incomplete data in 2 articles [36,38], pretrial and posttrial design in 1 article [42], 1 article being a study protocol [39], 1 N-of-1 trial [43], and incompatible data units in 2 studies [35,40]. Ultimately, 3 articles [37,41,44] were included in the meta-analysis.

The studies included in this review were published between 2015 and 2022 [35-44], and they were conducted in various countries, predominantly in the United States (n=6), Canada (n=2), and Spain (n=2). Participants’ ages ranged from 55 to 89 years, with 4 studies focusing on individuals aged ≥60 years, and 2 studies on those aged ≥55 years. The sample sizes of the included studies varied from 8 to 160 participants. A total of 8 studies included male and female participants, with one study focused exclusively on women and another exclusively on men (Table 1).

Of the 10 included studies [35-44], interventions were conducted in community settings (n=3), participants’ homes (n=3), health care centers (n=2), clinics (n=1), and hospitals (n=1). These interventions ranged from a minimum duration of 25 days to a maximum of 6 months. The most common intervention frequency was once daily (n=7), while 4 studies used interventions at different time intervals, such as every 15, 20, or 30 minutes. In 2 studies, the interventions occurred weekly (n=2). One study recommended 150 minutes of moderate-to-vigorous physical activity per week. Another study conducted 1 session per week for 4 weeks, followed by monthly sessions for 5 months. In addition, one study [44] implemented the intervention 5 times per week for 30 minutes per session or 3 times per week for 20 minutes per session. The most common intervention duration was 12 weeks (n=4), followed by 3 months (n=2). Other intervention durations included 25 days, 4 weeks, 13 weeks, and 6 months. The studies were conducted in general populations (n=6), and others had a disease-specific focus (n=2). Objective sedentary behavior assessment tools, such as accelerometers, were used in 8 studies, while subjective sedentary behavior assessment tools, such as questionnaires, were used in 2. The primary outcomes measured were sedentary behavior in 3 studies. The interventions resulted in decreased sedentary behavior (n=3). Study designs varied across the included studies: 8 studies [35-41,44] used a 2-arm or 3-arm RCT design, 2 studies [38,42] used a pre-post trials design, and one study [43] used an N-of-1 trial design (Tables 1 and 2). The control conditions included no intervention (n=2), only health-related information (n=2), nutritional counseling (n=1), brief counseling, and informative leaflets (n=1; Table 2).

The included studies targeting sedentary behaviors in older adults used the latest technology, synchronizing MHAs on tablets and smartphones with wearable devices. These smart devices included smartphones (n=7), tablets (n=2), and smartwatches (n=1). For example, Ashe et al [35] used smartwatches to record step counts, which were then used to calculate the increases in steps during training sessions for older adults. Lyons et al [37] and Li et al [42] synchronized MHAs on tablets with fitness bands to monitor sedentary behavior in older adults. In terms of control groups, there were various protocols implemented across different studies as shown in Table 2.

Among the 10 studies included in this review, 5 [35,38,41-43] extensively applied various theories to their mHealth intervention studies by focusing on sedentary behavior and physical activity among older adults. However, behavior change theoretical frameworks in mHealth interventions for sedentary behavior among older adults remain underused. These primarily include the habit formation theory [43], self-efficacy theory [42], the social cognitive theory [35], and the social-ecological theory [35] (Table 3).

This review synthesized studies using BCTs in MHAs to improve sedentary behavior among older adults. A total of 6 studies [35-37,41,42,44] incorporated BCTs, while 4 did not. The categorization primarily guided by the BCT taxonomy v1 developed by Michie et al [23,45] was applied to 5 studies [35-37,41,42], while one study [44] used the CALO-RE taxonomy [46]. Lyons et al [37] used the most BCTs (n=22) such as goal setting (behavior), problem-solving, action planning, review behavior goal(s), discrepancy between current behavior and goal, feedback on behavior, and social support (unspecified). Mackey et al [44] implemented 14 BCTs in their interventions, such as action planning, barrier identification and problem-solving, feedback on performance, and social support planning. Ashe et al [35] used 13 BCTs, while Blair et al [36] used 12, with both studies focusing primarily on the BCTs of goals and planning and repetition and substitution. Li et al [42] and Rosenberg et al [41] used fewer BCTs (n=6 and n=5, respectively), with both studies incorporating the following 5 BCTs: goal setting, feedback on behavior, self-monitoring of behavior, social support, and prompts/cues. Details of the BCTs used and the number of BCT interventions in 6 studies can be found in Multimedia Appendix 3.

The Cochrane risk-of-bias assessment results for the RCT studies are presented in Figure 2A and 2B; 8 of 9 studies [35-39,41,43,44] demonstrated a low risk of bias in all assessed domains (randomization process, deviations from intended interventions, missing outcome data, measurement of the outcome, and selection of the reported result). One study [40] showed concerns regarding the randomization process and overall risk of bias, while maintaining a low risk in the other domains. All 9 studies [35-39,41-44] indicated no significant issues in these 5 key areas, demonstrating a low risk of bias overall. The MINORS score for the nonrandomized study [42] was 11 (range 0-16), indicating a study of moderate methodological quality. Detailed MINORS score is provided in Multimedia Appendix 3.

Overall, 3 studies [37,41,44], involving sitting time interventions among 153 older adults were included in the meta-analysis. A fixed-effects meta-analysis revealed a statistically significant decrease in sitting time among older adults receiving mHealth interventions compared with those receiving conventional health interventions or no intervention (WMD=59.1, 95% CI 99.1 to 20.2; Z=3.0; P=.003; Figure 3).

This systematic review and meta-analysis evaluated the effectiveness of mHealth interventions in reducing older adults’ sedentary behavior. The systematic review showed that most interventions were delivered via mobile devices in community and home settings, were primarily focused on increasing physical activity, and were implemented once daily. The meta-analysis found a significant pooled estimate of 59.1 minutes per day (95% CI 99.1 to 20.2), indicating that mHealth interventions can notably decrease sedentary behavior in older adults. However, variations in intervention settings, frequency, duration, target outcomes, and other factors might affect the consistency and generalizability of these findings. Furthermore, some studies either did not use BCTs or applied very few and lacked a systematic theoretical framework. Notably, half of the included studies were conducted in the past 5 years, reflecting a growing trend in such interventions. Nevertheless, despite their high methodological quality, the studies had relatively small sample sizes and lacked representation from developing countries, as all included studies were conducted in high-income countries.

A total of 3 studies [37,41,44] in the meta-analysis had common features: using iPads or smartphones, having interventions lasting for 15 minutes or 1 hour, promoting recommended physical activity goals, lasting for 12 weeks, applying objective assessment tools, and including BCTs related to goals and planning, feedback and monitoring, social support, and associations. Overall, 2 studies used iPads [37,44], while one used a smartphone [41], suggesting that both device types are effective. In 2 studies [37,41], daily interventions were conducted, with 1-hour or 20-minute intervals for mobile prompts to disrupt sedentary behavior. One study [44] recommended at least 150 minutes of weekly moderate-to-vigorous physical activity without specifying the distribution of these minutes. Thus, intervention frequencies of 20-minute and 1-hour intervals, or similar activity-promoting goals, may be effective. Twelve-week interventions demonstrated significant effects; only one study [35] explored the effect over 6 months. All 3 studies [37,41,44] used objective assessment tools. Two studies [37,41] applied ActivePAL, which is considered the gold standard for measuring sitting time [47], and one used ActiGraph GT3X+ [44]. Being more sensitive to sedentary behavior changes than accelerometers, ActivePAL is widely regarded as one of the most effective measurement tools [43,48-50]. Thus, the sedentary time data in these studies were relatively accurate.

Three studies [37,41,44] were conducted in diverse settings: a home [37], a community [44], and a hospital [41], in all of which sedentary behavior was effectively reduced. Only one study [41] used a theoretical framework with sedentary behavior as the primary target; the other 2 [37,44] focused on physical activity. A previous study [51] showed that interventions solely targeting sedentary behavior were more effective than combined interventions. Notably, many studies [35-40,42,44] concluded that sedentary behavior and physical activity interventions often reported sedentary behavior as a secondary outcome. An analysis of 3 studies on mHealth intervention found that replacing sedentary behavior with physical activity as a means of reducing sedentary behavior did not significantly reduce sitting time in older adults [36]; this was in line with another study [41]. Nevertheless, 2 RCTs [41,43] using mHealth to interrupt sitting time in older adults achieved positive results, highlighting the importance of targeted sedentary behavior reduction interventions.

Furthermore, all 3 studies [37,41,44] integrated multiple BCTs related to goals, planning, feedback, monitoring, social support, and associations. Multicomponent BCT interventions generally yield better results than single-technique interventions in sedentary behavior reduction [51]. Core techniques such as goal setting, self-monitoring, and feedback were part of an effective intervention associated with sedentary time reduction, as supported by existing studies [37,52]. mHealth tools, with real-time monitoring, can prompt older adults to take action during prolonged sedentary periods [37,52]. They also feature persuasive elements, such as personalized motivational messages and progress tracking, to maintain engagement [53]. With highly customizable features, mHealth can be tailored to older adults’ needs and preferences, enhancing the intervention’s effectiveness [54]. This comprehensive approach makes mHealth a potentially more effective tool for reducing sedentary behavior compared with other digital interventions [18].

Previous meta-analyses [55-57] emphasized the advantages of theory-based interventions, yet few specifically addressed sitting time. In studies targeting sedentary behavior, Rosenberg et al [41,43] highlighted the significance of leveraging theories such as the social ecological theory, social cognitive theory, and habit formation theory to facilitate behavior change. Sedentary behavior is shaped by multiple factors, including physiological, psychological, social, environmental, and policy aspects [58]. The behavior change wheel (BCW) [45] should be applied in interventions, particularly those targeting both sedentary behavior and physical activity. The BCW offers a structured framework for designing behavior change interventions. Its 3-layer wheel model focuses on capability, opportunity, and motivation (inner layer), 9 intervention functions (middle layer), and 7 policy categories (outer layer) [52]. The BCW considers individual characteristics (behavior sources), intervention methods (functions), and social factors (policy categories) as crucial for intervention success [52]. Michie et al [52] proposed 93 BCTs, grouped into 16 categories based on BCW, to facilitate an insightful understanding of behavior change mechanisms and more precise tailoring of interventions. A systematic review showcased that theory-based, multi-BCTs are most effective for health behavior change [59]. A previous study, including one using BCW [60], analyzed sedentary behavior mechanisms and designed interventions for occupational populations and confirmed the effectiveness of BCTs in reducing sedentary behavior and the practical value of BCW.

A meta-analysis of relatively healthy community-dwelling older adults [18] showed that mHealth interventions might reduce sedentary time and promote physical activity in the short term (≤3 months). However, nonsignificant results and the inclusion of high-risk studies based on ROB 1.0. Similarly, another meta-analysis with 21 trials [61] had limitations, with only one study on older adults and combined eHealth interventions, potentially diluting the impact of mHealth interventions on sedentary behavior. Our study addresses these limitations in several ways. First, we expanded intervention settings beyond the community to include home, clinical, and health care center settings. Second, we included older adults with a broader range of health conditions such as cancer survivor and obesity, not just healthy ones. Third, we used the revised Cochrane ROB 2.0 tool, with all included studies demonstrating low bias risk, which is more rigorous than using ROB 1.0 before. Finally, our study focused solely on older adults. Therefore, by isolating mHealth interventions, our study provided more substantial evidence of the effectiveness of mHealth in reducing older adults’ sedentary behavior.

Overall, future intervention guidelines should be multifaceted to effectively address sedentary behavior in older adults. First, leveraging the BCW could help in comprehensively understanding and tackling the determinants of sedentary behavior. Second, mHealth intervention goals should be tailored to older adults, focusing on altering sedentary behavior and incorporating BCTs related to goals and planning, feedback and monitoring, social support, and associations. Thirdly, exploring different BCT combinations’ effects on reducing sedentary behavior is essential for promoting and maintaining positive behavior change. Fourthly, interventions could use devices such as iPads or smartphones, incorporate daily prompts at intervals of 20 minutes to 1 hour, and last at least 12 weeks, with extended follow-up durations to ensure effectiveness and sustainability. Furthermore, future studies, particularly in developing countries, should focus on specific target populations, explore different intervention durations, and assess long-term effects. Finally, comprehensive measurement of specific sedentary behavior in older adults should be emphasized, alongside the standardization of measurement units (eg, minutes per day, percentage of the day, and hours per week) to improve the universality and accuracy of research findings.

This study has certain limitations. Firstly, it only includes articles published in English, which limits the global comprehensiveness of our research outcomes. Secondly, the lack of uniformity in measurement units used by these tools presents challenges in conversion. Thirdly, most existing studies focus on combining physical activity with sedentary behavior, predominantly emphasizing the augmentation of physical activity. Literature directly addressing the reduction of sedentary behavior in older adults is relatively rare, making it challenging to ascertain the distinct role of reducing sedentary behavior. Finally, since this field is rapidly evolving, new studies may have been published after our search cutoff date, potentially influencing the current evidence base.

This systematic review and meta-analysis evaluated mHealth interventions for managing sedentary behavior in older adults. The findings indicate that mHealth interventions significantly reduce sedentary behavior in this demographic. Given the potential of mHealth intervention strategies that incorporate specific frequencies, durations, dedicated mobile monitoring devices, and tailored BCTs, these findings underscore the critical role of mHealth in translating research into effective public health strategies for older adults.