Effectiveness of interventions to address different types of vulnerabilities in community‐dwelling older adults: An umbrella review

Abstract Background Frailty, social isolation, loneliness, and poverty may render older adults vulnerable to social or health stressors. It is imperative to identify effective interventions to address them especially in the context of COVID‐19 pandemic. Objective To identify effective community‐based interventions to address frailty, social isolation, loneliness, and poverty among community‐dwelling older adults. Design Umbrella review. Data Source We systematically searched PubMed, Ovid MEDLINE, Embase, Cochrane CENTRAL, EBM‐Reviews, CINAHL via EBSCO, and APA PsycInfo via Ovid from January 2009 to December 2022. Eligibility Criteria We included systematic reviews or quantitative reviews of non‐pharmacologic interventions targeting community‐dwelling older adults. Data Selection, Extraction, and Management Two review authors independently screened the titles and abstracts, performed data extraction and appraised the methodological quality of the reviews. We used a narrative synthesis approach to summarize and interpret the findings. We assessed the methodological quality of the studies using AMSTAR 2.0 tool. Results We identified 27 reviews incorporating 372 unique primary studies that met our inclusion criteria. Ten of the reviews included studies conducted in low‐middle‐income countries. Twelve reviews (46%, 12/26) included interventions that addressed frailty. Seventeen reviews (65%, 17/26) included interventions that addressed either social isolation or loneliness. Eighteen reviews included studies with single component interventions, while 23 reviews included studies with multi‐component interventions. Interventions including protein supplementation combined with physical activity may improve outcomes including frailty status, grip strength, and body weight. Physical activity alone or in combination with diet may prevent frailty. Additionally, physical activity may improve social functioning and interventions using digital technologies may decrease social isolation and loneliness. We did not find any review of interventions addressing poverty among older adults. We also noted that few reviews addressed multiple vulnerabilities within the same study, specifically addressed vulnerability among ethnic and sexual minority groups, or examined interventions that engaged communities and adapted programs to local needs. Conclusion Evidence from reviews support diets, physical activity, and digital technologies to improve frailty, social isolation or loneliness. However, interventions examined were primarily conducted under optimal conditions. There is a need for further interventions in community settings and conducted under real world settings in older adults living with multiple vulnerabilities.

1 | PLAIN LANGUAGE SUMMARY 1.1 | Diets, exercise and digital approaches may improve frailty, social isolation or loneliness in older adults Exercise alone and in combination with diet may prevent frailty and improve social functioning in community-dwelling older adults.
Similarly, digital tools like connecting with others through the Internet may be effective in reducing loneliness. However, there is no clear evidence that such programmes are effective in groups including the LGBTQ2 + community and ethnic minorities.

| What is this review of reviews about?
Social isolation, loneliness and frailty are a serious public health risks that that affect many older adults, specifically people living in their own homes. Several interventions have been proposed to reduce these vulnerabilities. However, the effectiveness of these interventions is inconsistent in the general population, and unknown in specific populations.

What is the aim of this review?
This review of reviews summarises evidence on the effectiveness of interventions aimed at improving social isolation, loneliness and frailty among older adults. It also identifies gaps in evidence where further systematic review evidence is needed.

| What studies are included?
We included 27 reviews that were comprised of 372 unique primary studies. The vast majority of the reviews included studies that were conducted in high-income countries. Most reviews talked about either social isolation or loneliness.
Half of the reviews included studies with simple interventions, while the other half were more complicated, with many components.
Many of the studies had important weaknesses.

| What are the main findings of this review?
Systematic reviews suggest that exercise combined with nutritional supplementation have the highest odds of decreasing frailty, compared to nutritional supplementation of proteins alone at 3-4 months of follow-up.
Similarly, grip strength significantly improves when participants exercise and take protein supplements. Physical activity interventions also improve social functioning and reduce social isolation and loneliness.
There is a lot of conflicting evidence and inadequate reporting of results to determine effectiveness.
We were unable to find studies that looked at minority groups.

| What do the findings of this review mean?
Even though there is evidence in support of some interventions, only a small number of reviews systematically compared effects of interventions on social isolation and loneliness.
More studies are needed addressing other vulnerable groups or older adults living with vulnerabilities. This would allow for more definitive recommendations regarding the effectiveness of interventions for reducing frailty, social isolation and loneliness.
1.6 | How up-to-date is this review?
The search was conducted up to December 2022.

| BACKGROUND
Population ageing is increasingly prevalent across the globe, starting in high-income countries, and now in low-and middleincome countries (LMICs) (World Health Organization, 2021). By 2050, two-thirds of the world's population over 60 years of age will live in LMICs (WHO, 2022). Ageing results in biological changes that gradually deteriorate physical and mental capacities that increases the risk of disease and mortality. In turn, older age is characterized by the emergence of several complex health states commonly called geriatric syndromes. They are often the consequence of multiple underlying factors including frailty, urinary incontinence, falls, and pressure ulcers. These conditions decline the quality of life and the ability of older adults to contribute to their families and communities.
Adverse health outcomes in older adults are influenced not just by biomedical factors but also by vulnerabilities that could manifest in multiple ways (Department of Economic and Social Affairs programme on ageing, 2020; Chen & Schulz, 2016;Hoogendijk et al., 2019;Luanaigh & Lawlor, 2008;Waddell et al., 2018). Frailty defined as a state which affects individuals who experiences an accumulation of deficits in physical, psychological, and social domains leading to worsened adverse outcomes such as mortality and disability (Rockwood & Mitnitski, 2007). Similarly, social isolation, which is known as the subjective feeling of isolation, lack of belonging (Ekwall et al., 2005), loneliness, which is the subjective perception of feeling alone and absence of social network (Prohaska et al., 2020;World Health Organization, 2023), and minimal social engagement are prevalent among older adults and are all associated with a broad range of adverse outcomes. Financial vulnerability could further impede the ability of older adults to manage their illnesses and further threaten their wellbeing. Furthermore, infection prevention measures such as physical distancing and lockdowns implemented during the COVID-19 pandemic have further increased the likelihood of older adults experiencing new vulnerabilities or amplified pre-existing ones (Sepúlveda-Loyola et al., 2020).
Social isolation, loneliness, poverty, and frailty are independent concepts but related to each other (Yanguas et al., 2018). Social isolation and loneliness are associated with an increased likelihood of physical frailty among older adults (Gale et al., 2018;Makizako et al., 2018). Evidence suggest a bidirectional relationship between social isolation, loneliness and frailty (Mehrabi & Béland, 2020).
Systematic review evidence also suggests a strong relationship between poverty and frailty among older adults (Hayajneh & Rababa, 2021). Individuals whom have experienced a transition to poverty and remained in poverty were at a higher risk of frailty (Watts et al., 2019;Youn et al., 2020). In addition, social isolation among other psychological factors, may explain the relationship between poverty and frailty, supporting the notion of poverty as a complex of interlinked risk factors (Stolz et al., 2017). Furthermore, community-dwelling older adults are more likely to experience poverty and other types of vulnerabilities, especially with advancing age (Adepoju et al., 2021;Kaplan & Berkman, 2019).
There is strong evidence for the ability of individual physical activity, psychosocial and educational interventions to improve health outcomes in older adults (Sepúlveda-Loyola et al., 2020). Physical activity programs have been shown to improve reduce falls, improve strength, walking performance, and balance outcomes among frail older adults (Brady et al., 2020;Di Lorito et al., 2021). Social engagement and group activity programs have shown improvements in mental health and emotional well-being (Fancourt & Steptoe, 2018;Thomson et al., 2018;Noice et al., 2014). However, the effectiveness of the intervention varies according to their characteristics and theory underlying their intervention design (Di Lorito et al., 2021;Dickens et al., 2011). In contrast, the effectiveness of home visiting and befriending schemes is unclear (Cattan et al., 2005). The wide range of interventions and inconsistent findings on their effectiveness indicate a need for a comprehensive summarization and synthesis of the literature. Furthermore, given the interplay between different vulnerabilities, identifying strategies that address at least two of the three vulnerabilities (social isolation, loneliness, frailty) will help develop multi-component programs that mitigate their effects among older adults.
In conducting a rapid scan of the literature and review of the Campbell and Cochrane databases, we were unable to identify reviews of reviews focused on interventions to mitigate the consequences or alleviate vulnerabilities affecting older adults.
Therefore, the objectives of this umbrella review are to (1) identify available systematic review evidence on the effects of interventions in improve social isolation, loneliness, and frailty among communitydwelling older adults, (2) identify differences in effects of interventions across gender and ethnic minorities, (3) assess gaps in evidence where further systematic review evidence is needed.

| METHODS
This study is part of the Connection New Brunswick project, a project that aims to improve health and social well-being of vulnerable seniors in four New Brunswick communities through targeted codesigned programs based on identified vulnerabilities. We created an interdisciplinary committee to lead the review process, meetings (five formal meetings) were conducted to define the key domains of vulnerability, to establish a rigorous method and to validate the analytical approach. We report this review according to the Joanna Briggs Institute's guidance for the conduct and reporting of an umbrella review (Aromataris et al., 2015). We registered the protocol in PROSPERO under the number CRD42020190419.

| Eligibility criteria
We based our eligibility criteria on the characteristics of the participants of the primary studies included in the reviews, the nature of the interventions, the type of review and the reported outcomes. ADEKPEDJOU ET AL. | 3 of 34

| Participants
We included reviews where participants of the primary studies have the following characteristics: (a) at least 50% of the participants were 65 years or over, OR the mean age of the participants was at least 65 years, OR the age cut-off was not mentioned but the review referred to older adults as the target population, OR the age cut-off was 50 years and above but the review referred to older adults as the target population; (b) participants were community-dwelling older adults, (c) participants were healthy or had health conditions including chronic diseases and comorbidities.
We included reviews that provided findings for community-dwelling adults separately from older adults living in institutional settings. We excluded reviews where the setting of the older adults was not clear.

| Intervention(s), exposure(s)
Non-pharmacologic community-based interventions that aimed to address frailty, social isolation, loneliness or poverty among older adults. We excluded reviews studying pharmacotherapy only.

| Comparator(s)/control
The comparator was any other non-pharmacological intervention, usual care, no intervention, or placebo intervention.

| Types of review
We included any review (e.g., systematic, scoping, rapid) with or without a meta-analysis that met the following definition (Krnic Martinic et al., 2019): (a) aims to synthesize all the empirical evidence that meets pre-specified eligibility criteria to answer a specific research question (b) used systematic methods to identify, appraise and synthesize empirical evidence, (c) provided an explicit and reproducible search strategy of at least one or more bibliographic databases (d) synthesized the findings descriptively or quantitatively. Reviews must have included at least one randomized controlled trial (RCT) or non-randomized study of interventions (NRS) (e.g., quasi-experimental studies, controlled clinical trials, before and after studies/pre-post design, Interrupted Time Series).
We excluded narrative reviews developed without systematic literature search, review protocols and reviews that were not in English or French given the language capacity of the team members.

| Outcomes
We included reviews where frailty status, social isolation or loneliness were identified as primary or secondary outcomes regardless of how they were measured (validated or non-validated instruments). We also included reviews that reported at least three of the five criteria of Fried's frailty phenotype (weakness, slow gait speed, low physical activity, exhaustion, unintentional weight loss).
For social isolation we considered social isolation and its related dimensions such as social functioning, social participation, social activities, social support, social network size, social connectedness.
For this umbrella review, our primary outcomes were measures of improvement in frailty status, Fried's frailty phenotype components (weakness, slow gait speed, low physical activity, exhaustion, and unintentional weight loss), social isolation, loneliness, and poverty. We have selected those outcomes due to evidence of their interrelationship (Yanguas et al., 2018). Client-centered outcomes examined were improvements or the decrease in risk of adverse health outcomes (e.g., depression, falls, death), health service use (e.g., hospitalization, premature admission to long-term care) and costs. These outcomes reflect longer-term real-world consequences of interventions. which were developed using an iterative process to minimize false positives and optimize results, are described in Supporting Information 1.

| Search strategy
We searched for relevant systematic reviews in the reference list of the included systematic reviews.

| Study selection
Two review authors independently screened the titles and abstracts generated by the search against the inclusion criteria. We retrieved and screened the full-text reports for all titles that appeared to meet the inclusion criteria in duplicate and independently. We resolved disagreement through discussion or by seeking the opinion of another review author. We documented the reasons for excluding reviews. Neither of the review authors was blind to the journal information, the study authors, or institutions.

| Data extraction
Using a standardized data extraction form (in Excel sheets), two reviewers extracted data from the reviews independently. To ensure consistency across reviewers, we conducted calibration exercises before starting the review to develop an extraction guide. Data abstracted included first author's name, year of publication, review's objective (s), type of included studies, the number of included studies, details on participants (sample sizes, age, gender, disease or condition), details on intervention (name, type of deliverer, setting, modes of delivery, duration), details on comparison (name, type of deliverer), outcome measures (frailty, social isolation, loneliness, poverty), type of analysis (qualitative synthesis, meta-analysis), main results, name of the interventions that have a positive effect, duration of follow-up, the assessment of risk of bias of the primary studies as reported by the review authors and the evaluation of the certainty of evidence as reported by the review authors. Reviewers resolved disagreements by discussion, or by a third reviewer. When the information provided in the reviews was not clear, we went back to the primary studies included in the selected review to extract these data.

BOX 1:
Checklist for critical appraisal of the methodological quality of systematic reviews (AMSTAR 2) (Shea et al., 2017)

| Methodological quality assessment
Two independent reviewers appraised the methodological quality of the retained reviews using the AMSTAR 2 (Assessing the Methodological Quality of Systematic Reviews) critical appraisal tool (Shea et al., 2017). Items covered by this appraisal are presented in Box 1.
Disagreements were resolved first by discussion and then by consulting one of the team members for arbitration.
We considered seven methodological quality items as critical. These included item 2 (Protocol registered before commencement of the review), item 4 (Adequacy of the literature search), item 7 (Justification for excluding individual studies), item 9 (Risk of bias from individual studies being included in the review), item 11 (Appropriateness of meta-analytical methods), item 13 (Consideration of risk of bias when interpreting the results of the review), item 15 (Assessment of presence and likely impact of publication bias). We considered the remaining items as noncritical. We judged not meeting a critical methodological quality item as a critical flaw. We judged not meeting a non-critical methodological quality item as a non-critical weakness. Based on the approach proposed by Shea and colleagues, we judged the overall confidence in the result of a review as high (no or one non-critical weakness), moderate (more than one non-critical weakness), low (one critical flaw with or without non-critical weaknesses) or critically low (more than one critical flaw with or without non-critical weaknesses) (Shea et al., 2017). We did not evaluate the certainty of evidence of the overview using GRADE because it has not been adapted to reviews of reviews. Instead, we collected and reported the certainty of evidence evaluated by the authors of the included reviews.

| Strategy for data synthesis
We tabulated data related to characteristics of included reviews. We tabulated the description of the interventions identified in included ADEKPEDJOU ET AL. | 5 of 34 reviews, the number of participants (from included primary studies), the effect of the interventions and their magnitude as reported in the published reports. We summarized the effects of the interventions descriptively using tables. We also tabulated the assessment of risk of bias of the included primary studies and the evaluation of the quality of the evidence as reported by the review authors. Finally, we mapped the overlapping primary studies contained within the included reviews.
We explored the effect of the interventions within the following subgroups: interventions primarily focusing on either frailty, social isolation, loneliness, adverse health outcomes or costs separately. Reviews that reported multiple outcomes of interest were presented accordingly. We presented separately the results of systematic reviews from other type of reviews (e.g., scoping reviews). We also presented the results of unimodal interventions (e.g., physical activity alone) and multimodal interventions (e.g., interventions combining diet and physical activity) separately. Within each intervention, we distinguished the results of randomized controlled trials from the results of other study designs. We planned to explore the effectiveness of the interventions within the following subgroups: high-income versus LMICs, urban versus rural setting, men versus women.

| Characteristics of included reviews
Twenty of the included reviews were systematic reviews and seven were narrative reviews with systematic search or scoping reviews (Anton et al., 2018;Hagan et al., 2014;Ibrahim et al., 2022;Kelaiditi et al., 2014;Pool et al., 2017;Puts et al., 2017;Wister et al., 2021).
Seven of the systematic reviews conducted a meta-analysis (Burton et al., 2019;Frost et al., 2017;Fu et al., 2022;Li et al., 2022;Liao et al., 2018;Shvedko et al., 2018;Walters et al., 2017). Walters and colleagues meta-analyzed RCTs and NRSs assessing home-based health interventions separately . Frost et al. (2017). metaanalyzed the same set of RCTs. Therefore, to avoid duplicate reporting of findings, we only reported the meta-analysis findings of the NRS from Walters et al., not their RCT findings . A Systematic review on social prescribing for older adults conducted by Smith et al.
(2019) did not find any eligible studies but was included because it met the eligibility criteria. The 26 reviews included 372 unique primary studies. A total of 64 primary studies were included in multiple reviews (Supporting Information 1). Ten of the reviews included studies conducted in LMICs according to the World bank classification.

| Characteristics of the primary studies included in the reviews
The number of eligible RCTs per review ranged from 3 (Pool et al., 2017) to 47 (Theou et al., 2011). The number of eligible nonrandomized studies (NRS) per review ranged from 2 (Puts et al., 2017) to 24 (Khosravi et al., 2016). RCTs included parallel RCTs, pragmatic RCTs, cluster RCTs, randomized crossover trials.
NRSs included a variety of designs: single group pre-and post-test studies/before-and-after studies, quasi-experimental pre-and post-test studies with control group, pilot/feasibility/exploratory studies, evaluation/post-hoc evaluation studies, cohort studies, case-control studies, cross-sectional studies, and case studies.
Twenty-five reviews reported on studies sample sizes. The sample size ranged from 3 (Khosravi et al., 2016)
c This was meta-analysis of RCT, not observational studies. Since these RCT were also included in Frost et al. (2017); we only considered the observational studies to avoid duplicate.
T A B L E 2 Effect of non-pharmacological strategies to address frailty status among community-dwelling older adults.

Interventions
Author  -1 RCT: the intervention group compared to control group had significantly greater improvements in 2 of the 3 measures of frailty (the modified PPT, the peak oxygen uptake, but not in ADL function).
Not reported -1 RCT: At 12 months the prevalence of frailty in the intervention group was significantly lower than the control group.
10% vs. 19.1% (p < 0.05) At 12 months the mean number of frailty markers decreased significantly more in the intervention group compared to the control group. The total frailty score after intervention was lower in the exercise group compared to control group.
7.1 (SD 4.0) vs. 8.0 (SD 4.8) (p < 0.001) During the year, the percentage of older adults newly certified for LTCI (considered frail) was lower in the intervention group compared to control group.

months a significant difference between intervention and
Not reported control group in the reduction in the number of frail persons.
-1 RCT: 58% (Exercise + milk fat globule membrane) vs. 28% (milk fat globule membrane) (p < 0.05) The percentage of non-frail participants at postintervention was significantly higher in the [Exercise + milk fat globule membrane group] than in the milk fat globule membrane group or placebo. Similarly, the mean Edmonton Frail scale score also significantly improved in intervention group compared to the control group) Abbreviations: CI, confidence interval; HR, hazard rate; OR, odds ratio; RCT, randomized controlled trial; SMD, standardized mean difference.
T A B L E 3 Effect of non-pharmacological strategies to address social isolation among community-dwelling older adults. Due to the lack of available data, meta-analysis for social isolation outcomes was not performed.   -1 before-and-after study: following mentoring service and community activities (one-to-one intervention), there were significant improvements in social support at 12 months (p = 0.02) -1 NRS (n = 320): No significant differences between experimental a learning control groups for social participation.
-2 NRS (n = 103): Significant improvement in social integration after the intervention.
Note: Significant between-group differences are shown in bold. Therapeutic social interventions include: Community Connections' group, Mindfulness Based Stress Reduction Program, Day center attendance, LUSTRE 6-week group program (social program focusing on positive self-management and wellbeing), Friendship enrichment program, Psychosocial group intervention, Senior Companion Program Befriending scheme (one-to-one intervention), Community-based mentoring Service (one-to-one intervention), Mentoring service + community activities (one-to-one intervention).
Abbreviations: CI, confidence interval; CT, controlled trial; NS, no significant results; RCT, randomized controlled trial; SD, standard deviation; SMD, standardized mean difference.  1 non-randomized controlled trial (n = 140): The intervention group showed a non-significant decrease in loneliness when comparing baseline to 6-months follow-up. -

Educational activity
Cohen-Mansfield and Perach (2015) 12 trials (n = 2796) -1 randomized (for one intervention group only), controlled trial: Both forms of intervention resulted in significantly less loneliness in a group by time interaction analysis. Follow-ups were 2-and 12-month postintervention.

Not reported
-1 not randomized controlled trial: At postintervention, the difference in reduction of loneliness between the experimental and control group was significant.

Not reported
-1 not randomized, not controlled trial: Participants experienced significant reductions in loneliness at postintervention compared to baseline. The greatest drop in loneliness was seen in low-income ethnic minorities and minorities with high levels of education.
Not reported -6 RCTs: no significant between groups differences -2 not randomized controlled trial: no significant between groups differences --1 not randomized, not controlled trial: no significant between groups differences -Physical activity interventions Shvedko et al. (2018) Due to the lack of available data, meta-analysis for loneliness was not performed 1 RCT (n = 32): no significant between groups differences -Cohen-Mansfield and Perach (2015) 1 RCT with 2 intervention (exercise) groups (n = 174): In both conditions, loneliness significantly decreased at the end of the intervention but significantly increased at 6-month follow-up. Intervention group mean reduced from 42.35 to 37.40, compared with the control group's increase from 38.40 to 40.75; (p = 0.008) -3 controlled trials or RCTs: no significant between groups differences -1 pilot study (n = 764): this pilot study could not adequately evaluate the programme's potential impact on loneliness. -

T A B L E 4 (Continued)
Interventions Authors Effect on loneliness Magnitude of the effect 1 cross-sectional study (n = 817): no statistically significant differences between 417 day-centre attendees and 400 older people who did not attend in terms of their reported loneliness.
-1 evaluation (mixed methodology) (n = 66): examination of a Senior Companion Programme (Befriending scheme, one-to-one intervention), which matched volunteers with older individuals, measured loneliness on one occasion and reported a below average scoring for the participants (a mean of 31, the author stating that a typical score for those between 60 and 80 years of age being between 32 and 37) that did not find any eligible studies), 24 reported on duration of interventions. the minimum duration was 1 week (Khosravi et al., 2016) and the maximum duration was 120 months (Ibrahim et al., 2022) (Table 1). Two reviews did not report on duration of interventions (Hagan et al., 2014;Looman et al., 2019). There was great variety in the measurement tools used to assess social isolation and its related dimensions. Instruments included:
Note: Significant between-group differences are shown in bold. Therapeutic social interventions include: Community Connections' group, Mindfulness Based Stress Reduction Program, Day center attendance, LUSTRE 6-week group program (social program focusing on positive self-management and wellbeing), Friendship enrichment program, Psychosocial group intervention, Senior Companion Program Befriending scheme (one-to-one intervention), Community-based mentoring Service (one-to-one intervention), Mentoring service + community activities (one-to-one intervention).
None of the reviews reported on the psychometric characteristics of the instruments (Supporting Information 4).

| Interventions addressing frailty
In Table 2  Regarding factors associated with frailty status, evidence from observational studies shown that normal and overweight Body Mass Index, higher Mini Mental State Examination (MMSE) score were protective against worsening frailty or predicted greater likelihood of improvement from prefrail (prefrail being defined as presenting one or two of the five criteria of Fried's frailty phenotype) to robust among prefrail men . Higher Mediterranean diet (MED) score was associated with lower risk of frailty whereas food insufficiency (albumin, selenium, and carotenoids) and lower vitamin D levels were associated with higher risk of frailty or prefrailty (presenting one or two of the five criteria of Fried's frailty phenotype) (Kelaiditi et al., 2014).
Regarding walking speed, the results from systematic reviews and other type of reviews were conflicting and did not provide Regarding physical activity, a systematic review reported that the two included trials (n = 501) that tested physical activity interventions shown an improvement of this outcome following the intervention (Theou et al., 2011). The results from the other reviews did not provide enough evidence in favor of an improvement of physical activity following the other types of intervention.

| Interventions addressing social isolation
A systematic review with meta-analysis reported a small significant positive effect in favor of physical activity interventions on social functioning (SMD = 0.30; 95% CI = 0.12-0.49; p = 0.001, I 2 = 63%, 22 RCTs, n = 1625) (Shvedko et al., 2018). Another systematic review reported that three of four included trials (n = 412) and two longitudinal surveys (n = 3095) shown a reduction of social isolation, an increase of social inclusion and an increase of satisfaction in the amount of contact with others following Internet use (Khosravi et al., 2016). The results from the other reviews were conflicting or ADEKPEDJOU ET AL.
| 27 of 34 did not provide enough evidence in favor of an improvement of social isolation following the other types of intervention (Table 3).

| Interventions addressing loneliness
A network meta-analysis of 3 RCTs and 15 NRS (n = 1523) on physical activity and psychosocial interventions from found an insignificant improvement in loneliness (Hedge's g = 1.09; 95% CI = 0.57 to 1.62) (Li et al., 2022). Similarly, a meta-analysis of four RCTs (n = 165) of remotely delivered interventions aimed at reducing loneliness found no significant effect (p > 0.05; SMD = 0.18 [95% CI = −0.03 to 0.40]; I 2 < 50%). The results from the other reviews were conflicting or did not provide enough evidence in favor of an improvement of loneliness following the other types of intervention (Table 4).

| Effect of interventions on adverse health outcomes, health services use and cost
Eight reviews (Arantes et al., 2009;Burton et al., 2019;Coll-Planas et al., 2017;Dedeyne et al., 2017;Franck et al., 2016;Ibrahim et al., 2022;Looman et al., 2019;Theou et al., 2011) reported on the effect of the interventions on adverse health outcomes and cost. The results from these reviews were conflicting or did not provide sufficient information to describe the effects (e.g., no effect size or statistical measures) of the interventions on adverse health outcomes or health services use (Supporting Information 7).
Regarding costs, a systematic review reported that a physical activity intervention helped reduce the expenditure on institutionalization and medical visits (one trial, n = 104) (Arantes et al., 2009). Another systematic review reported that physical activity reduced paid caregiver support (1 RCT, n = 186) (Burton et al., 2019). Ibrahim et al (Ibrahim et al., 2022) reported that group interventions significantly reduced healthcare costs

| Subgroup analyses
We did not find any reviews that conducted subgroup analyses between participants of LMICs and high-income countries, individuals living in urban settings versus rural settings, and men versus women.

| Methodological quality assessment of the reviews
The quality of the reviews ranged from low to critically low indicating that all the included reviews had at least one critical flaw. All the reviews included the components of PICO in the research questions, described the inclusion criteria and the selection of the study designs for inclusion, and used a comprehensive literature search strategy.

| Methodological quality of the primary studies in the reviews
Eighteen of the 21 systematic reviews assessed the methodological quality of the included studies. For two systematic reviews, it was unclear whether the authors assessed the quality or not (Cohen-Mansfield & Perach, 2015;Khosravi et al., 2016). One systematic review did not assess the quality because there were no studies found that met the eligibility criteria (Smith et al., 2019). Two of the other type of reviews assessed the methodological quality (Pool et al., 2017;Puts et al., 2017). Cochrane risk of bias tools were the most frequently used by the reviews to assess methodological quality of the included studies (Burton et al., 2019;Coll-Planas et al., 2017;Frost et al., 2017;Fu et al., 2022;Looman et al., 2019;Shvedko et al., 2018;Sims-Gould et al., 2017;Tricco et al., 2022). Other tools were: the PEDro scale (Arantes et al., 2009;Liao et al., 2018), the Downs and Black checklist for methodological quality assessment of randomized and nonrandomized studies of healthcare interventions (Franck et al., 2016), the methodological index for nonrandomized studies (MINORS) (Dedeyne et al., 2017), the Jadad methodological quality criteria scale (Dedeyne et al., 2017), the Newcastle-Ottawa Scale (Pool et al., 2017;Walters et al., 2017) (Li et al., 2022). For most of the reviews, the quality of the included studies was rated as moderate to high (Arantes et al., 2009;Burton et al., 2019;Dedeyne et al., 2017;Franck et al., 2016;Heins et al., 2021;Liao et al., 2018;Looman et al., 2019;Pool et al., 2017;Puts et al., 2017;Shvedko et al., 2018;Tricco et al., 2022;Theou et al., 2011;Walters et al., 2017). For four reviews, most of the studies had high or unclear risk of bias (Coll-Planas et al., 2017;Frost et al., 2017;Fu et al., 2022;Li et al., 2022)

| DISCUSSION
In this umbrella review, we identified systematic reviews that suggested that programs focused on protein supplementation combined with physical activity may improve frailty status, and frailty components like grip strength and body weight. They also suggested that physical activity alone or in combination with diet may prevent frailty and may improve social functioning.
The quality of the reviews ranged from low to critically low which is comparable to other assessments of reviews of interventions addressing geriatric care (Conneely et al., 2022;Lozano-Montoya et al., 2017). Furthermore, several evidence gaps within the reviews have also become apparent such as: effects of multi-component interventions focused on older adults with multiple vulnerabilities, effects of community-focused interventions adapted to local needs, or interventions that impacted client-centered outcomes. Moreover, we did not identify reviews summarizing effects of interventions designed to mitigate the effects of poverty or focused on marginalized communities in LMICs.
The results of this umbrella review are consistent with the literature. An umbrella review of seven systematic reviews including 58 randomized trials found a benefit of resistance training and multicomponent exercise interventions that include resistance, aerobic, balance and flexibility tasks (Jadczak et al., 2018). An overview of 10 systematic reviews found a benefit of exercise combined with amino acid supplementation (Lozano-Montoya et al., 2017). A recently published large-scale longitudinal cohort study found that exercise therapy was associated with a greater likelihood of improvement in frailty status among older adults receiving home care services (Larsen et al., 2020). Despite its observational nature, this study involving 250,428 people from several jurisdictions provides real world evidence of potential benefit of community exercise programs.
The fact that physical activity may improve social functioning (social vulnerability) in addition to frailty (health vulnerability) is critical. Indeed, many older adults may fall into more than one vulnerability category (Johnson & Wiener, 2006;Lee et al., 2018).
Given the interrelationship and potential negative interactions between health, social and financial vulnerability, the interventions that target only one vulnerability category may fail to address its underlying determinants and may not be effective in the longterm. Therefore, physical activity could be an essential component of multimodal health promotion interventions that aim to address more than one vulnerability category. Nevertheless, more primary research is needed to identify other effective strategies to address more than one vulnerability category within the same study population.
Regarding technology interventions, a previous review of systematic reviews found that Internet-supported interpersonal communication such as videoconferencing shown evidence of effectiveness in reducing loneliness (Chipps et al., 2017). This is in line with our results that suggest that Internet use may improve social isolation and loneliness. A recently published evidence and gap map documented the many intervention categories used in digital interventions (Campbell Collaboration; Welch et al., 2022). This review again documented the lack of equity considerations in this field of study.
This umbrella review revealed additional areas of knowledge gaps. First, published reviews did not identify many large-scale community-based real-world interventions that engaged communities and adapted programs to local needs. One review (Hagan et al., 2014) reported a complex intervention, the "Community Connections," a community-based program designed to engage and mobilize resources for older adults in effective interactions with younger populations. However, the authors of the review commented that this primary study was done on a small scale and was not able to adequately evaluate the program's potential impact on loneliness. Second, there was no comment on the location of primary studies within individual reviews. Indeed, our scan of primary studies within reviews appeared to all be from middle-and high-income countries and no studies were conducted in low-income countries. We did identify one published trial of a peer-to-peer support pilot intervention to improve quality of life among highly vulnerable, low-income older adults in Cape Town, South Africa. This study demonstrated an increase of social interactions and physical activities and a decrease of loneliness after volunteers visits with clients. This very recent primary study was not included in the reviews we selected but does provide evidence that these interventions are feasible in low and middle income settings (Geffen et al., 2019).
Due to methodological flaws, the overall quality of the reviews ranges from low to critically low. This is primarily due to the absence ADEKPEDJOU ET AL.
| 29 of 34 of prespecified protocol, the absence of a list of excluded studies with justification of their exclusion and the lack of satisfactory technique for assessing the risk of bias (RoB) of included observational studies. Although reporting guidelines and quality assessment tools exist, adherence to reporting guidelines and quality assessment tools are not consistent across systematic reviews Pussegoda et al., 2017;Shamseer et al., 2015). Some evidence is emerging that biases within systematic reviews could influence results and quality of overviews of systematic reviews (Page et al., 2014). Therefore, mechanisms to improve adherence to established reporting guidelines and methodological assessment tools are needed to improve the quality of systematic reviews.
Regarding generalizability, for the majority of the reviews, populations included in the primary studies' suffered from several health and/or social issues such as prefrailty, frailty, receiving home care services, having disability, cancer, geriatric syndromes, chronic diseases, anxiety, depression, mental health conditions, social conditions (social isolation, loneliness), or diverse health conditions. However, participants' sociodemographic were not captured or reported within individual reviews with one exception.
Only one review aimed to identify effective interventions to address social isolation and loneliness in community-dwelling older adults of ethnic minority groups. This review found that volunteering activity, educational activity, physical activity and physical activity combined with educational activity may improve social isolation and loneliness. However, these results were based on a small number of studies and all the included studies were done in the United States (Pool et al., 2017). In addition, none of the included reviews addressed the impact of the interventions on gender. While there are a number of studies and reviews that focus on loneliness and/or social isolation among older adults in the general population as well as known male and female differences in loneliness, there appear to be few sex/gender sensitive interventions designed to target it (Maes et al., 2019;Rapid Response Service, 2020). Because of the victimization and discrimination toward them, many lesbians, gay, bisexual, and transgender, queer and 2-spirit (LGBTQ2+) older adults are vulnerable to social isolation or loneliness.
LGBTQ2 + older adults are more likely to live alone and less likely to have children as their non-LGBTQ2+ peers. This further puts this population at a higher risk of loneliness and social isolation and requires specific attention (Disparities, 2011;Goldsen, 2018). Therefore, the results of this umbrella review could apply to community-dwelling older adults presenting health and social conditions, but more research is needed to identify what strategies are effective among, financially vulnerable older adults, ethnic and gender minorities.
The strengths of this umbrella review include the comprehensiveness of the search strategy, the recency of the reviews and the large number of primary studies and participants representing the effects interventions on frailty, social isolation and loneliness in community-dwelling older adults. It is also the first umbrella review that was fully designed to address more than one area of vulnerability. A limitation of this umbrella review is restriction of inclusion to English and French languages. The Four studies that were excluded might have contained relevant evidence. It is possible that there are other recent reviews published in other languages that were not captured (Naito et al., 2021). However, we believe it would be unlikely given that social isolation and loneliness are primarily observed in high-income countries and manuscripts are usually published in English language. Another limitation is the lack of consensus in the definition of frailty and the use of various criteria to define frailty. Given the growing body of research in this topic there is an urgent need for standardizing the measurement of frailty using frailty assessment scales. Finally, the umbrella review did not capture reviews that identified largescale community-based real-world interventions with community engagement and adaptation of programs to local needs. This may be explained by the fact that we addressed vulnerabilities rather than healthy aging outcomes and we focused on peer-reviewed publications. systems are in place to support the delivery of social and health promotion activities. We did not seek out the term "social prescribing." We also did not review core elements of such programs to ensure that they are successfully deliver on improved health and social outcomes.

| Recommendations for future research
In this umbrella review, we identified several knowledge gaps.
Specifically, studies aiming to mitigate multiple risk factors,