This study was registered in PROSPERO (International Prospective Register of Systematic Reviews; CRD42022346823). The PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines (Multimedia Appendix 1) were followed in this review [22].

Studies were included in this review if they met the following inclusion criteria:

Studies with measurement instruments including individual items on empowerment but not providing individual item scores were excluded.

A systematic search was carried out in the following electronic databases: PubMed, ScienceDirect, SCOPUS, EBSCO, and SciELO. The search strategy consisted of a combination of terms based on previous reviews on empowerment [24] and technology [25] (for details, see Multimedia Appendix 3). In PubMed, the following search string was used: ((“Empowerment”[Mesh] OR “Patient Participation”[Mesh]) OR “self-efficacy” OR “locus of control” OR perceived autonomy OR perception of autonomy OR overconfidence) AND ((“Technology”[Mesh]) OR “Digital Technology”[Mesh] OR “Mobile apps” OR m-health OR ehealth OR e-health OR “digital solution” OR “Ambient Assisted Living” OR “digital services” OR website OR “digital platform” OR esocial). This string was adapted for the remaining databases. No filters were used. The search was conducted on May 4, 2022, and updated on September 6, 2022, and included papers from each database’s inception until August 31, 2023.

After importing the papers into Mendeley (Elsevier), duplicates were removed. Next, the papers were imported to CADIMA (Julius Kühn-Institut), the software used to support screening, which was performed first by title and abstract and then by full text against the eligibility criteria. The entire selection process was performed independently by 2 authors (AIM and AGS). Disagreements or discrepancies during the selection process were resolved by consensus or consultation with a third author (NPR or OR).

A customized Microsoft Excel form was used for data extraction, which was piloted using 3 studies. The data extraction was manually completed by 1 author (AIM) and checked by a second author (AGS). The following data were extracted from each study: author and date, measurement instruments used to assess the outcomes of interest, intervention (type and duration), characteristics of the sample per group (health condition, sample size, age, sex), comparison, outcomes, and results (at postintervention and follow-up assessments). Data on the period of follow-up and the results of follow-up assessments were collected to allow for analysis of the effectiveness of digital solutions at different time points. When multiple outcomes within the same domain of empowerment were reported in the same paper, the choice of which to include in the meta-analysis was based on similarity with outcomes from other studies in the same meta-analysis. When data were not fully accessible (eg, presented in graphs), they were not extracted and authors were contacted.

Extracted data were summarized under 1 of the 4 domains of empowerment previously defined (patients’ capacities, patients’ behaviors, patients’ knowledge, and support by others). When necessary, the transformation of data, such as the conversion of SEs and CIs into SDs, was performed according to Cochrane recommendations [26].

The quality of studies was assessed using the PEDro (Physiotherapy Evidence Database) scale, which is a valid instrument [27]. Each study was assessed independently by pairs of reviewers (AIM and AGS, NPR and OR), and disagreements were resolved by a third author. The PEDro scale is composed of 11 items, and each item is scored as either present (1) or absent (0). The first item is not considered in the total score, leading to a maximum score of 10. The total score can be interpreted as follows: 0-3, poor; 4-5, fair; 6-8, good; and 9-10, excellent methodological quality [28].

The certainty of evidence was graded for the main comparisons and outcomes of this review. The overall quality and strength of the evidence per outcome was assessed according to the GRADE (Grading of Recommendations Assessment, Development, and Evaluation) framework. Evidence was assessed against (1) risk of bias, (2) inconsistency of results, (3) indirectness, (4) imprecision, and (5) publication bias, and the overall quality of evidence was classified as high, moderate, low, or very low [29,30]. We started with high evidence as studies included in the meta-analysis were randomized trials, and then, the evidence was downgraded, considering the researchers’ assessment of each criterion.

When at least 3 studies were available within the same empowerment domain and comparison group (no intervention, usual care, or face-to-face intervention), a meta-analysis was performed. In addition, separate analyses were performed considering whether the digital intervention was the only intervention or was administered together with a face-to-face intervention. A sensitivity analysis was performed excluding lower-quality studies. Effect sizes (ESs) and their 95% CIs were determined by the standardized mean difference (SMD) and classified according to Cohen’s guidelines as small (0.20), medium (0.50), and large (0.80) effects [31]. Statistical heterogeneity was investigated using I² statistics interpreted as 25%, 50%, and 75%, reflecting low, moderate, and high statistical heterogeneity, respectively. When at least 10 studies were available, funnel plots were generated to inform publication bias. Follow-up periods were chosen to increase the similarity between studies in the same meta-analysis. The meta-analysis was performed using R version 4.2.2 (R Foundation for Statistical Computing). When no meta-analysis was possible, a qualitative synthesis was performed, indicating the comparisons investigated by the studies and the direction of the effect reported.

A total of 10,875 references were found, of which 946 (8.7%) were duplicates. The remaining 9929 (91.3%) references were screened by title and abstract, and 301 (3%) full-text manuscripts were retrieved and read. Of these, 30 (10%) manuscripts were included in the review (Figure 1). The included manuscripts assessed the impact of using a variety of digital solutions, including the following: websites (n=6, 20%), general mobile apps (n=4, 13.3%), condition-specific mobile apps (n=6, 20%), desktop software (n=1, 3.3%), remote monitoring (n=6, 20%), remote monitoring plus a website (n=1, 3.3%), remote monitoring plus mobile apps (n=5, 16.7%), and remote monitoring, a specific mobile app, plus a website (n=1, 3.3%).

The studies scored between 3 and 8 on the PEDro scale. Most studies scored 6-8 (n=20, 66.7%), suggesting good quality; 9 (30%) studies scored between 4 and 5, suggesting fair quality; and 1 (3.3%) study scored 3, suggesting poor quality (Multimedia Appendix 3). The certainty of evidence varied between very low to low and is presented next, along with the results of each meta-analysis (Multimedia Appendix 4).

Data extracted from individual studies are presented in Multimedia Appendix 5.

This review synthesized existing evidence on the effectiveness of using integrated digital solutions for empowering older adults. Results suggest with very-low-to-low certainty of evidence that using digital solutions results in increased knowledge and increased capacities when compared to usual care and impacts capacities to an extent similar to face-to-face interventions. Interestingly, results also suggest, with low certainty of evidence, that there are no differences between using digital solutions and no intervention for improving capacities.

Our results show an apparent contradiction as they suggest that digital solutions are better than usual care and equal to face-to-face interventions at improving patients’ capacities but equal to no intervention. This might be explained by the characteristics of usual care, which, in the studies included, seems to equate to minimal care delivered over 1 or few sessions. As previously reported [62,63], usual care is generally not described in sufficient detail, hindering the interpretation of results. In addition, differences in the degree of interaction and support provided by the digital solutions, in the training given to individuals before the trial, in the amount of interaction with health professionals using the digital solutions, and in the personal and clinical characteristics of the participants might contribute to the differences found. Furthermore, the very-low-to-low certainty of the evidence, the wide CIs for the estimate of the effect, and the high heterogeneity in the usual-care meta-analyses suggest that further research is likely to impact the conclusions of the comparison of digital interventions against no intervention and usual care. Worth highlighting are the 2 studies that compared digital solutions against face-to-face interventions at follow-up and that confirmed the postintervention results by showing no between-group differences. Most studies within the capacity domain assessed self-efficacy (ie, people’s beliefs in their own capabilities), which is reduced in older adults receiving care [64] and is an important determinant of health-related behavior change [65].

Digital interventions seem to improve older adults’ knowledge to a higher extent than usual care, which, as previously reported, was not detailed in the included studies. The instruments used to assess acquired knowledge are diverse, but most relate to knowledge regarding a specific clinical condition. Other systematic reviews have reported that digital solutions can improve one’s knowledge, for example, related to diabetes [66], and are better than usual care for eHealth literacy [67]. A previous study also reported that using digital solutions for seeking health-related information (eg, on specific diseases, medication, symptoms, and health promotion) is common among older adults [21], potentially facilitating adherence to digital interventions that promote knowledge acquisition.

This systematic review with meta-analysis results carefully suggests that the choice of using a digital solution or not might be left to individual preferences and that technology can be used to reach more individuals who would not have access to face-to-face empowerment strategies. Other studies have identified patients’ needs and preferences as important aspects to consider when deciding between in-person health care delivery or remote health care delivery [68,69]. In addition, the decision on whether to use a digital solution might depend on the domain of empowerment being specifically targeted as relevant for a particular individual, potentially choosing digital solutions to target knowledge and capacities and face-to-face interventions for other domains. Choosing a mixed approach where some domains of empowerment are targeted with face-to-face interventions, while others are targeted with digital interventions might also be an option. In addition, issues of security and privacy should be considered when choosing digital solutions for older adults, as they often have lower levels of digital literacy and consequently are more exposed to privacy violations, frauds, scams, and phishing attacks [70,71]. Developing digital solutions that have user-friendly interfaces and implementing privacy by design (ie, incorporating privacy features into the design of digital solutions, such as data minimization principles) are essential. Additionally, education and awareness are a means to minimize the risk and to empower older adults with the knowledge and skills to identify online risks [70].

The similarity of results found in this review for the domains of patients’ knowledge and capacities might suggest that increased knowledge might promote increased capacities by providing older adults with more knowledge about their health, resources, and management strategies. The study of causality was beyond the scope of this review. This result, rather, identifies a potential topic of study for future studies.

Regarding patients’ behaviors and support by others, it was not possible to aggregate the evidence in a meta-analysis due to insufficient studies with a similar comparison. The qualitative analysis of the studies included in our systematic review identified conflicting results regarding the impact of digital interventions. The factors previously identified as potential sources of heterogeneity might also contribute to the conflicting results.

Future studies should detail the intervention administered to the control group, particularly if it is usual care, as well as detail the functionalities of the digital solution provided, including training, interaction with professionals, and adherence to the intervention, as these factors might have an impact on the effectiveness of the digital intervention to facilitate empowerment. Future systematic reviews and meta-analyses should attempt to identify sources of heterogeneity that might impact the results, including the type of digital solution, the amount of support given to participants, participants’ characteristics (age, education, digital literacy), the length of the intervention, and subdomains within each domain of empowerment.

The studies included are highly diverse in terms of the type of digital solutions, health conditions, participants’ characteristics, and instruments used to assess empowerment. All these aspects are reflected in the high heterogeneity found and undermine the confidence in the results of the meta-analysis. We aimed to run a subgroup meta-analysis considering the type of technology and users’ characteristics, but the small number of studies precluded this. For some comparisons, a meta-analysis was not possible due to the reduced number of studies with a similar comparison. This might suggest that empowerment is more researched when related to health domains. Studies that were not based on empowerment theory were not included.

There is very-low-to-low certainty of evidence that using integrated digital solutions results in increased knowledge and increased capacities (mainly self-efficacy) when compared to usual care and that it impacts capacities to an extent similar to face-to-face interventions at postintervention. Interestingly, results also suggest, with low certainty of evidence, that there are no differences between using digital solutions and no intervention for improving capacities. Included studies and technologies are diverse, and the meta-analysis showed high heterogeneity, which limits the confidence in the results and suggests that further research might affect the conclusions of this review. The limited number of studies precluded a subgroup analysis considering the type of technology and user characteristics.