Article Text

PDF

Social capital interventions targeting older people and their impact on health: a systematic review
  1. Laura Coll-Planas1,2,
  2. Fredrica Nyqvist3,
  3. Teresa Puig2,4,
  4. Gerard Urrútia2,5,
  5. Ivan Solà2,5,
  6. Rosa Monteserín2,6
  1. 1Fundació Salut i Envelliment (Foundation on Health and Ageing), Universitat Autònoma de Barcelona, Barcelona, Spain
  2. 2Institute of Biomedical Research (IIB Sant Pau), Barcelona, Spain
  3. 3Faculty of Education and Welfare Studies, Social Policy, Åbo Akademi University, Vaasa, Finland
  4. 4Servicio de Epidemiología Clínica y Salud Pública, Hospital de la Santa Creu i Sant Pau, Universitat Autònoma de Barcelona, Barcelona, Spain
  5. 5CIBER de Epidemiología y Salud Pública (CIBERESP), Madrid, Spain
  6. 6Equip d'Atenció Primària Sardenya, EAP Sardenya, Barcelona, Spain
  1. Correspondence to Dr Laura Coll-Planas, Fundació Salut i Envelliment UAB, Casa Convalescència UAB, C/ Sant Antoni M. Claret 171, 4a planta Barcelona 08041, Spain; laura.coll{at}uab.cat

Abstract

Background Observational studies show that social capital is a protective health factor. Therefore, we aim to assess the currently unclear health impact of social capital interventions targeting older adults.

Methods We conducted a systematic review based on a logic model. Studies published between January 1980 and July 2015 were retrieved from MEDLINE, EMBASE, CINAHL, PsycINFO, Cochrane Central Register of Controlled Trials and Web of Science. We included randomised controlled trials targeting participants over 60 years old and focused on social capital or its components (eg, social support and social participation). The comparison group should not promote social capital. We assessed risk of bias and impact on health outcomes and use of health-related resources applying a procedure from the Canadian Agency for Drugs and Technologies in Health (CADTH) based on vote-counting and standardised decision rules. The review protocol was registered in PROSPERO (reference number CRD42014015362).

Results We examined 17 341 abstracts and included 73 papers reporting 36 trials. Trials were clinically and methodologically diverse and reported positive effects in different contexts, populations and interventions across multiple subjective and objective measures. According to sufficiently reported outcomes, social capital interventions showed mixed effects on quality of life, well-being and self-perceived health and were generally ineffective on loneliness, mood and mortality. Eight trials with high quality showed favourable impacts on overall, mental and physical health, mortality and use of health-related resources.

Conclusions Our review highlights the lack of evidence and the diversity among trials, while supporting the potential of social capital interventions to reach comprehensive health effects in older adults.

  • AGEING
  • SOCIAL CAPITAL
  • HEALTH IMPACT ASSESSMENT
  • RANDOMISED TRIALS
  • SYSTEMATIC REVIEWS

Statistics from Altmetric.com

Introduction

Societal and structural changes are reframing social contacts in quantity and quality. Among older people, the risk of social isolation and loneliness is increasing, while opportunities for social relationships and participation might emerge. Accordingly, the WHO Active Ageing paradigm highlights that social networks shape resilience and health throughout life.1 ,2

Social capital has several definitions.3 From the social cohesion approach Putnam refers to it as a public good based on community activities.4 This approach is the most widespread in health research and was adapted to the ageing process emphasising the interaction between individuals at the individual (or micro) and the collective levels, comprising meso (neighbourhood) and macro (society) contexts.5 ,6 Accordingly, we use social capital to refer to an umbrella concept, in which social resources (social capital components) are grouped into dimensions: social networks, social contacts and participation belonging to the structural or objective aspects; and social support, sense of belonging and trust corresponding to the cognitive or subjective aspects. Moreover, depending on the directions of social ties, social capital is defined as bonding (intragroup ties between members sharing common characteristics), bridging (ties between heterogeneous groups) or linking (relationship between people who possess unequal wealth, power and status).7–9

Observational studies indicate that social capital components are a major protective factor for mental and physical health and mortality, with an effect comparable to smoking cessation.10–13 Moreover, theoretical background and empirical evidence show how specific dimensions and directions of social capital are related to specific gains and losses of health, as well as to health inequalities.14–16 Social capital interventions, specially those that promote social support and social participation, have often the purpose to increase well-being or mental health, alleviate loneliness, promote healthy lifestyles or improve self-management of chronic diseases. However, it remains unclear whether social capital interventions impact on the variety of health outcomes linked to social capital according to observational studies. Some trials achieved significant effects on several health outcomes while others have not,17–20 and an overview of trials that promote social capital is lacking. Previous systematic reviews have generally included non-randomised designs21–24 and focused interventions on specific social capital components25 ,26 or psychosocial interventions.24 ,27 Moreover, they have assessed psychosocial effects and seldom considered health outcomes.22

Furthermore, social capital also generates undesirable consequences, which are understudied.28

Social capital interventions are complex and, consequently, pose specific challenges, for example, regarding impact mechanisms and implementation.29 Moreover, the lack of evidence hampers the implementation of social capital interventions in healthcare. Accordingly, an evidence base for further research, policy and practice is essential.

Therefore, we conducted a systematic review of the literature, broad in scope, with the objective of assessing the impact on health outcomes and use of health-related resources of interventions that promote social capital or its components among older people.

Methods

A review protocol was developed and registered in PROSPERO (reference number CRD42014015362). We report the results according to the PRISMA statement.30

We considered eligible those studies with a randomised controlled trial design, that included participants over the age of 60 (or alternatively with a mean age over 64). Studies had to assess an intervention that promoted social capital or one of its components.6 In multicomponent trials, the inclusion was restricted to those studies in which social capital was the focus of the intervention. Professional support was not considered social support, and thus was not social capital either.31 We included studies reporting effects on health outcomes (general, physical and mental health or mortality) or use of health-related resources (including nursing home placement). Comparison groups could not contain social capital components.

We conducted an exhaustive search of articles published between January 1980 and July 2015 in: MEDLINE, EMBASE, CINAHL, PsycINFO, the Cochrane Central Register of Controlled Trials and the Web of Science.

We combined a series of text terms and controlled vocabulary related with the population and the intervention of interest. We added to this algorithm an adaptation of the Cochrane filter to identify controlled trials. No language restrictions were applied. We include the complete search algorithms in online supplementary appendix 1. References of study protocols, systematic reviews and included studies were checked for additional studies, and we contacted the first authors from the included studies.

Supplementary appendix 1

Two review authors (LC and RM) independently screened the results retrieved from the search to check eligibility criteria. We obtained the full text of eligible studies and independently assessed their final inclusion. Discrepancies were resolved by consensus or by consulting with a third author (FN).

We designed a data extraction form to obtain data from included studies and describe their characteristics in terms of design, population, context, intervention, comparison, outcomes and results.

We described participants according to the disadvantage categories from the PROGRESS Plus framework, considering that social capital is a social determinant of health that is highly interrelated with the rest of the determinants.32 ,33

We classified the interventions according to the social capital dimensions, the directions of the social ties and whether promoted relationships were new and/or existing. We extracted data according to the TIDieR reporting guidelines to characterise the programmes described.34

We adapted the Cochrane risk of bias tool to assess the internal validity rating random sequence generation, allocation concealment, blinding and incomplete outcome data.35 We report a summary judgement on risk of bias for each study, according to random sequence generation, allocation concealment, and incomplete outcome data rated as: ‘high risk of bias’ when at least one domain was determined to be biased, ‘low risk of bias’ if all domains were rated as unbiased, and ‘unclear’ when at least one item was not reported in detail to make judgements. Blinding was excluded from the summary risk of bias due to its difficulty to be implemented in social capital interventions.

We contacted study authors for missing data in the trial reports.

We could not perform a quantitative synthesis using meta-analysis due to the clinical diversity in terms of participants' characteristics, intervention designs, settings and contexts, outcomes and measurement procedures.35 Indeed, similar reviews on psychosocial interventions have seldom found the proper conditions to apply meta-analysis.22 ,24 Moreover, the heterogeneity of ways of reporting results among the included studies prevented us also from comparing effect sizes and analysing whether the effects were clinically meaningful. Therefore, we conducted a narrative synthesis based on described effects to assess health impact.36

First, we identified sources of complexity and according to the ESRC (Economic and Social Research Council) guidance on the conduct of narrative synthesis,37 we built a logic model to support the conceptualisation outlining that social capital-based interventions might improve health outcomes and use of health-related resources by promoting physiological, psychological, behavioural and instrumental changes (see figure 1).38–42

Figure 1

Logic model illustrating the conceptual approach.

For the preliminary synthesis, we tabulated the information on study design, context (setting, geographical and policy context), target population, social capital-based intervention characteristics, social capital goals and components and health outcomes and use of health-related resources at study level. Undesirable outcomes were listed and classified.

In the next step, we clustered studies assessing the same outcome (eg, quality of life) and applied the standardised decision rules and statements about effectiveness to produce a narrative evidence synthesis used by the Canadian Agency for Drugs and Technologies in Health (CADTH) in the Rx for Change database.43 ,44 Accordingly, we limited the synthesis to outcomes that were reported in at least five trials, as in fewer studies the applicability of findings would be spurious. Outcomes insufficiently reported (ie, reported in less than five studies) were grouped for descriptive purposes into wider health categories, for example, physical health, and subcategories of subjective and objective outcomes were established when applicable. For each sufficiently reported outcome (ie, reported in at least five studies), we applied vote-counting to count the number of trials according to the reported direction of effect (significant favouring social capital, significant favouring control, non-significant differences). Afterwards, we assessed the impact of the interventions according to the following decision rules: ‘no effect’ if any included study favoured the intervention; ‘generally ineffective’ when up to 33% of the studies favoured the intervention; ‘mixed effects’ when 34–66% of the studies favoured the intervention; ‘generally effective’ when more than 66% of the studies showed a favourable effect. We completed the analysis relating these results with the directions and dimensions of social capital addressed in the interventions, the range of intervention length, the populations and settings targeted and the summary judgement on risk of bias. The accumulated sample size of trials was considered in order to weight the results with a descriptive purpose.

Finally, we focused on studies judged as low risk of bias and identified in which outcomes and outcome categories trials reported positive impacts.

Along the analysis, we differentiated effects on subjective and objective outcomes.

Results

Description of included studies

We screened 17 341 abstracts and included 73 papers reporting 36 randomised controlled trials. The eligibility process is described in a PRISMA flow chart (figure 2).

We summarise the trial characteristics in table 1 and provide the detailed information at study level in online supplementary appendix 2 as tabulated for the preliminary synthesis.

Supplementary appendix 2

Table 1

Basic descriptive table on the included studies

Trials were very heterogeneous regarding population, intervention characteristics, context and outcomes according to their measure, report and follow-up period. However, they were more frequently conducted in the community, in urban areas of high-income countries and mainly targeted Caucasian older people without disability or dementia.

The studies provided limited information on context. Six studies stated how specific policies supported their research: Active Ageing, national health priorities, policies on voluntary action, guidelines on specific diseases and research priorities.

Regarding the disadvantage categories,33 almost half of the studies had a mean age of between 70 and 80 years and 25% of 80 and over. Women were majority in 29 trials and were the exclusive target in three studies. Men were majority in only one study.45 In caregiver studies, women were majority among caregivers and men among care recipients. The category sexual orientation was not mentioned in any study. Eleven studies included ethnic minorities; in two most participants were African-American.20 ,46

Twenty-three studies reported participants' economic and/or educational level, but with heterogeneous descriptions. Ten studies mainly included people with low socio-economic levels. The lowest educational level was reported in a trial in which 47.1% of the participants had no primary education,47 while the highest was described in a trial in which 96% of participants had completed high school.48

Interventions ranged from 1.5 months to over 1 year. Around half of the interventions had a duration of 3 months or less and the last postintervention assessment reported was just after the intervention. Programmes were mainly based on social support (eg, support groups, peer support…), social activities, befriending schemes and/or engaging participants in activities. From the social capital perspective, the cognitive dimension and the new and bonding relationships were the most frequently promoted.

Interventions were delivered face-to-face in 28 studies, four were remote and four combined both modes. Volunteers, peers, students, lay workers and a wide range of health and social care professionals were involved.

Fifteen studies assessed group interventions. Some specificities to highlight are the use of a seal robot to promote social interaction in a nursing home and that two programmes promoted a health behaviour change. Some groups were remote.

Fourteen studies focused on individual interventions, mainly based on home visits or visits to the nursing home but two programmes were remote. Remarkably, one was a cognitive stimulation based on social interaction through computer. Three interventions involved members of the existing support network.

Three interventions combined individual with group-based activities.

Four studies applied a setting approach (ie, involving the complete institution), three programmes were based on intergenerational activities with schools and one provided humour therapy in a nursing home. No interventions were community-wide.

Regarding undesirable consequences of interventions, seven (19.4%) studies either reported harmful effects, mechanisms to detect them, or how they were solved. Precisely, four studies reported not having caused adverse mental events. Three further studies reported miscommunication, interpersonal friction and dissatisfaction with closure of the groups, and due to lack of face-to-face contact and shared interests. The first two adverse effects were solved during the intervention.

Risk of bias assessment is presented in figure 3. Only eight studies (22%) were considered to have an overall low risk of bias, while the majority was judged as unclear due to lack of reporting. The attrition rate ranged from 0% to 64.3%. Most studies (25) had an attrition rate below 25%. Attrition was equivalent among groups in 14 studies, but was higher in the intervention or the control group in seven studies each. Seventeen studies mentioned an intention-to-treat analysis, but only four explained how they imputed missing data.

Figure 3

Assessment of risk of bias. *The summary risk of bias is based on selection and attrition bias.

Only seven trials (19.4%) reported on blinding. Of those, four studies reported blinding the outcome assessors, but in one of them blinding was revealed, and one reported blinding of data analyst.

Effects on health outcomes and use of health-related resources

According to the CADTH methodology, quality of life, well-being, self-perceived health, mood (including depressive symptoms and anxiety) and loneliness were subjective outcomes sufficiently reported to be assessed, and mortality was the only objective outcome reported also in at least five trials. Mood was the most frequently studied outcome. Specifically, trials were interpreted as generally ineffective on loneliness, mood and mortality. Nevertheless, trials with successful results on those outcomes targeted complex cases of loneliness and depression.17 ,45 ,49 ,50 Also one trial with low risk of bias targeting lonely people was effective on mortality.18 Regarding quality of life, well-being and self-perceived health, trials reported mixed effects. Remarkably, some of those effective trials targeted lonely and depressed older people.51 ,52

Table 2 presents the narrative evidence synthesis on sufficiently reported outcomes and online supplementary appendix 3 details results at study level on those outcomes.

Supplementary appendix 3

Table 2

Narrative evidence synthesis on sufficiently reported outcomes

Table 2 shows the interpretation of the effectiveness and the applicability of the results of the sufficiently reported outcomes (ie, reported by at least five studies). It presents the range of length of the intervention and the aggregated sample sizes.

Regarding setting, positive effects were reported in community-dwelling older adults and nursing home residents in all sufficiently reported outcomes except for mortality.

Considering exclusively the studies judged as low risk of bias, favourable impact was reported in quality of life, well-being, self-perceived health, mood and mortality but not in loneliness.18 ,51 ,53–55

Regarding insufficiently reported outcomes, all categories had at least one positive outcome from a trial with low risk of bias: generativity, feeling needed, agitation in dementia and caregiver burden among psychological variables; physical activity and self-reported strength among subjective outcomes on physical health; walking speed, physical ability, aerobic fitness and percentage of body fat among objective outcomes on physical health; executive function, verbal learning, fluency and memory, and cortical and hippocampal volume among objective outcomes on cognition and intellectual activity among the subjective ones; visits to the doctor, days spent in hospital and nursing home placement among use of health-related resources. Online supplementary appendix 4 presents detailed results on these outcomes at study level with additional text.

Supplementary appendix 4

Discussion

Summary of findings

We identified 36 randomised trials assessing the health impact of a social capital intervention targeting older people.

Studies were clinically diverse but unequally distributed across settings and contexts, intervention designs, target population and outcomes assessed. Subjective outcomes were more frequently reported.

The harmful effects of social interventions were understudied, rare, mild, limited to mental health and, at least partially, resolvable.

According to the CADTH procedures, trials were generally ineffective on loneliness, mood and mortality and reported mixed effects on quality of life, well-being and self-perceived health. Nevertheless, those trials with successful results addressed complex cases or had low risk of bias. Moreover, in all sufficiently reported outcomes, but loneliness, at least one study with positive results had a low risk of bias. Therefore, our findings indicate the potential of social capital interventions to impact these outcomes.

In all categories of insufficiently reported outcomes (ie, psychological variables, physical health, cognition, use of health-related resources), at least one trial with a low risk of bias reported a positive impact, comprising subjective and objective outcomes.

In summary, although the review does not allow estimating the effect of the intervention, the narrative synthesis detected a signal that for certain populations and outcomes the intervention could be effective.

Strengths and weaknesses

This is the first systematic review of clinical trials focused on social capital targeting older people and assessing its health effects. Our results are consistent with previous reviews focused on specific social capital components, which show positive trends but inconsistent results and highlight the need for higher quality research.21–27

However, our review has several limitations; most of them linked to limitations of the available evidence.

The majority of studies were judged to be at high or unclear risk of bias. In addition to the lack of reporting of relevant details on methods and the limited scope for blinding, attrition was a high source of bias and intention-to-treat analysis was underused.

We applied the PROGRESS Plus framework.32 Although women were the main target, gender implications were seldom reported in background, intervention design and discussion. Many studies mentioned socio-economic status but heterogeneously. Disability and cognitive decline were frequently exclusion criteria. Several trials included minor ethnicities but rarely focused on minorities and seldom reported specific implications. Sexual orientation was not reported, despite its major consequences on support networks among older people.56 Moreover, contextual aspects were under-reported.

At review level, we managed following complexity sources: interventions with multiple components, relevance of contextual factors on implementation and outcomes, multiple outcomes of interest, and difficulty in locating, appraising and synthesising the evidence to answer the research question. The search strategy was complex and exhaustive and retrieved a high number of documents. Nevertheless, inconsistently labelled and poorly defined interventions might be difficult to locate. Furthermore, we developed a logic model to guide the review.

As in similar reviews,22 ,24 we could not perform a meta-analysis, or compare effect sizes and analyse whether the effects were clinically meaningful due to the mentioned diversity.35 Therefore, the standardised procedures from the CADTH based on vote-counting and decision rules were the best option as narrative evidence synthesis that allows to assess impact and interpret data in reviews with multiple outcomes and high diversity.43 Moreover, we combined sample sizes to weight results with a descriptive purpose. Nevertheless, a high number of pilot and small studies were probably underpowered to detect effects.

We applied a broad scope on health and thus identified a wide variety of subjective and objective measures such as physical and cognitive performance, blood and MRI parameters, health service use from medical records and data from mortality registers. Nevertheless, all objective outcomes except mortality were insufficiently reported to interpret effectiveness.

In this review, we addressed the heterogeneity of outcomes and focused on social capital as a whole including all components under the same concept. Moreover, we explored and described the frequency with which the interventions addressed the different directions and dimensions of social capital and to which outcomes were associated.

Interpretation of study results

Our results provide preliminary evidence that social capital might promote general health. However, they require cautious interpretation due to the high diversity and low quality of the trials. Impact on mood, loneliness and mortality may call for specific intervention designs. Indeed, social capital interventions seem to increase general health in lonely people although they do not relieve their loneliness.

Moreover, positive effects were reported in different contexts, participants' characteristics and intervention designs and in a wide variety of subjective and objective outcomes. Therefore, our review supports the potential of social capital to reach comprehensive health effects.

In addition, in our logic model we outlined four pathways that link social capital with health and these physiological, psychological, behavioural and instrumental intermediate outcomes were reported by some studies.

Specific trials showed less physiological damage through improvements on glycated haemoglobin, blood pressure, weight, BMI, waist circumference and percentage body fat.57–59

Psychological improvements were supported by trials with impact on self-esteem58 and caregiver burden,60–63 but effects on self-efficacy and mastery were not achieved.64 ,65

Behavioural changes were observed only regarding physical activity.59 ,66 ,67

Instrumental changes in terms of better health access were supported by one trial that successfully increased participation in cardiac rehabilitation.68

However, these intermediate outcomes were seldom analysed as mediators in the original trials,69 nor were their potentially mutual effects addressed.

Implications for practice, policy and research

First, evidence-informed policy has to be nurtured by research, but especially in the social capital field, health and social policy should be committed to contributing to the limited evidence by evaluating existing programmes, involving especially the third sector.

Furthermore, social capital interventions might contribute to reduce health inequalities by addressing social determinants of health.70 Accordingly, we encourage an inclusive approach when improving social capital by considering the disadvantage categories of the PROGRESS Plus framework in the design, evaluation and reporting.

Trials need to be conducted and reported applying quality standards,71 and need to use standardised health outcome measures including objective ones. Moreover, social capital interventions should be addressed in the frame of complex interventions.29 ,72 ,73

Further research should gain specific knowledge on subgroups of older people (eg, nursing home residents, caregivers and those suffering from chronic conditions).28 Loneliness, as a condition to target and as an outcome, regards special attention. A further focus should be on whether and, if so, how interventions based on different social capital dimensions and directions achieve differential health effects and contrast whether and how social outcomes mediate health changes.25 Moreover, the role of the length and intensity of the intervention and the type of relationship between the intervention and health effects (eg, linear, threshold) should be clarified. It is also important to address how to increase adherence and reduce attrition and how to establish mechanisms to detect, solve and report adverse events. Furthermore, the health impact of social capital interventions conducted at community level remains unknown,74 as well as how to tailor these interventions to different individual needs, cultures (eg, family-based vs individualistic) and welfare systems.6

Remarkably, these and further findings on effectiveness of social capital interventions should be carefully considered always in the frame of the specific purpose and value of the intervention. For instance, impact on mental health will be key factor on an intervention aimed at preventing depression, while lower glycated haemoglobin might be relevant when improving diabetes self-management.

Finally, as a next step, a taxonomy is being developed based on the results of the systematic review, aimed at structuring the diversity of social capital interventions to guide further research, policy and practice and thus potentially reach comprehensive health effects across older adult populations and contexts.

What is already known on this subject?

  • Numerous observational studies have shown that social capital resources are important for the understanding of health and well-being and isolated intervention studies based on social capital have achieved favourable results on health among older people. However, no systematic review of controlled trials has previously assessed the health impact of social capital interventions on older people.

What this study adds?

  • Our findings highlight the lack of evidence, high clinical diversity between trials and the low quality, while suggesting the potential of social capital to impact health, specially quality of life, well-being and self-perceived health in older adults. This review contributes towards building an evidence base for social capital interventions from a public health perspective to advance in the health and social care systems addressing social capital as a relevant protective health factor.

Acknowledgments

Laura Coll-Planas has conducted this study and published this paper within the PhD Programme of Preventive Medicine and Public Health at the Universitat Autònoma de Barcelona.

References

View Abstract

Footnotes

  • Twitter Follow Laura Coll-Planas at @EstudiAequalis and Rosa Monteserín at @rmonteserin

  • Contributors LC-P, RM and FN searched for, screened and selected studies. IS searched for studies. LC-P, RM and FN extracted data. LC-P, RM and FN conducted the analysis. All authors interpreted the analysis, drafted the final manuscript, and read and approved the final version. LC-P is the guarantor.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement All data used for the review are available from the authors.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.