Table 2.
Social capital conceptualizations and health outcomes.
| Reference | SC approach used to synthesize the results | SC indicators used to describe the results and show how the approach to SC is operationalized | Offered precise information on how SC was measured in each study | Key results | Overall relationship between SC and health (++, +, n.s., -) |
|---|---|---|---|---|---|
| Almedom (2005) | Cognitive and structural components of both bonding and bridging SC, on micro-(e.g. family SC) and macro-levels (e.g. community or neighborhood SC). | A broad variety of indicators used by the reviewed studies are reported. These are not grouped into SC dimensions. | Some precise information | There is consistent evidence that SC is linked to better mental health outcomes, but there are also negative relationships. The effects vary according to subgroups (low SES, minority groups, excluded groups, gender, life course) and contexts (e.g., SC can be harmful for health in disadvantaged neighborhoods). | +/− |
| Alvarez et al. (2017) | Family SC as conceptualized by cohesion and network approaches. SC dimensions are described as they are used in the reviewed studies; the authors did not re-group studies and indicators according to SC dimensions. |
Cohesion approach: Indicators for family cohesion: Collective efficacy, informal control, social interaction, sense of belonging Network approach: Indicators for family support: emotional support, instrumental support, family conflict; family network: network structure, quality of family ties. |
Precise information | There is consistent evidence that indicators of family SC in both the cohesion and network approaches were related to better health. Also, the mechanisms social influence/social control, social engagement and the exchange of social support can be related to better health. Yet, there was also evidence for non-significant relationships and the (so far still underexplored) negative impact of SC via mechanisms such as family conflict (e.g., conflicting goals or excessive demands), negative social support, and social influence. | ++/n.s./- |
| Carlson and Chamberlain (2003) | Broad category of community-level SC as conceptualized by Coleman and Putman; individual or geographic attributes, single or multi-level. | A variety of indicators are taken into account. The focus lies on perceptions of trust and social participation. | Some precise information | Most of the reviewed studies point to an association of SC and health, but the authors state that interpretation of the results is problematic due to a lack of conceptual and methodological development in the reviewed studies. There is also evidence for variations in the relationship between SC and health due to subgroups (black/white) and low/high income regions, but the mechanisms remain unclear. | + |
| Choi et al. (2014) | Seven dimensions of SC that are assessed at an individual- or area-level. These were adapted from the UK National Office for Statistics. | Seven dimensions of SC: Social participation, social network, civic participation, social support, trust, norm of reciprocity, sense of community. | Precise information | Looking at all levels, the results indicate lower mortality rates for higher levels of trust, social participation and civic participation. There were non-significant relations for social support, norm of reciprocity, sense of community and social networks. | I–S: +/n.s. I–C: +/n.s. |
| Coll-Planas et al. (2017) | Cognitive and structural components of SC at individual and community levels, bonding, bridging, linking SC, existing SC (relationships), and new SC (new relationships). | A broad variety of different indicators are assessed and are linked to SC dimensions. Indicators belong to the following categories: Social support, social participation, social network, trust and social cohesion. |
Some precise information | Although the majority of intervention studies failed to show a significant improvement of health outcomes, there is evidence for positive effects when considering low-risk-of bias studies and specific populations. | +/n.s./- |
| De Silva et al. (2005) | Cognitive and structural SC at both individual and ecological levels. | A broad variety of indicators that are used by the reviewed studies are reported. These are classified into either cognitive or structural SCs that can be assessed on either an individual or ecological level. Examples for cognitive indicators: Trust, social cohesion, sense of community. Examples for structural indicators: Group-membership, social participation. |
Some precise information | There is evidence that both cognitive and structural SCs are protective against the development of mental disorders at the individual level, but evidence is lacking or inconsistent on an ecological level. | I–C: ++ I–S: + I-mixed: + E-C: n.s. E-S: n.s./- E-mixed: +/n.s./- |
| Ehsan and De Silva (2015) | Cognitive and structural SC at both individual and ecological levels. | A broad variety of indicators that are used by the reviewed studies are reported. These are classified into either cognitive or structural SCs that can be assessed on either an individual or ecological level. Cognitive indicators: Trust, social cohesion, perceived social support, sense of community. Structural indicators: Group-membership, engagement in public affairs, actual social support, community networks |
Some precise information | Cognitive indicators of SC seem to be protective against the development of common mental disorder both on an individual and ecological level, whereas structural indicators point to no or mixed associations, with some evidence of negative associations in low resource settings. | I–C: ++ I–S: n.s./- E-C: ++; E-S: +/n.s./- |
| Flores et al. (2018) | Cognitive and structural SC or proxy indicators for these dimensions, both on an individual and ecological level. | A broad range of indicators are reported. Results are reported in an aggregated way according to the broader SC dimensions. Examples for structural indicators as given by the authors: group-memberships, social participation, social networks. Examples for cognitive indicators: group identification, social connectedness, and sense of community. | Precise information | Although most interventions improved health outcomes, evidence is inconsistent as to whether or not SC is improved. There is not enough evidence showing that an improvement in health outcomes is due to improvements in SC through the interventions. | I–C: +/n.s. I–S: +/n.s. E-mixed: + |
| Gilbert et al. (2013) | Different constructs of SC at either the individual level, the group/ecological level, or multi-level. | 102 measures were classified into constructs of SC: Participation, trust, efficacy, sense of community, social support, social networks, bonding (e.g., participation, trust, reciprocity), bridging (e.g. social networks, political or electoral participation), linking (e.g., voting and trust in legal, political, or government institutions), or combined (SC indices). | No precise information | The relationship between SC and health is on average positive; a one unit increase in SC is related to a 27% increase to be in good health (OR = 1.27; self-reported health: OR = 1.29, survival: OR = 1.17). Yet, there is significant heterogeneity that points to numerous negative and non-significant relationships. Results seem to vary due to the different constructs of SC, levels and countries, however, none of these heterogeneity tests reached significance. | ++/n.s./- |
| Hu et al. (2014) | Cognitive and structural SC on an individual or ecological level. | A broad range of indicators were assessed. Example indicators for cognitive SC given by the authors: trust, fairness, help, support, reciprocity; for structural SC: membership, participation, voting or election, homeownership, and crime rates. | Some precise information | There was good evidence that both cognitive and structural indicators (mixed levels) and individual- and ecological-level indicators (mixed dimensions of SC) were associated with less illness. However, there were also numerous non-significant relationships. There was consistent good evidence for SC as a protective factor against CVD (most strongly at the ecological level and for structural SC), diabetes (especially for cognitive SC), cancer (especially for structural SC) and no evidence in the few studies that investigated COPD. | I-mixed: +/n.s./- (CVD, diabetes) E-mixed: +/n.s. (CVD:++; cancer, diabetes: +; COPD: n.s.) C: +/n.s./- (CVD, diabetes) S: ++/n.s. (CVD, cancer; diabetes: n.s.) CVD: + Diabetes: + COPD: n.s. |
| Islam et al. (2006) | Individual level SC, aggregated SC (individual responses aggregated), or contextual SC, including multi-level. | A broad range of individual indicators and contextual indicators are taken into consideration when results are reported. These are grouped according to levels and study designs. The authors state, however, that most of their reviewed studies operationalized SC as a combination of both cognitive (particularly, trust and reciprocity) and structural (informal participation or civic engagement) dimensions, and at an area level, most studies used aggregated responses. | Precise information | There is strong evidence for a positive relationship between SC and health or survival at both individual and ecological levels. The positive association between SC and health varies between countries and is stronger in less egalitarian countries. Cross-level interactions indicate that living in an area with high SC is more strongly associated with health in less egalitarian countries. | Single-level mixed: ++ Multi-level mixed: ++ With moderators: ++/n.s./- |
| Khazaeian et al. (2017) | SC as one whole concept | Indicators for SC are not exhaustively listed. In the description of the results, the authors report results for three indicators: Trust, sense of belonging and social participation. | Some precise information | There is evidence for a positive relationship between the SC and the health of female heads of households. | I–C: + I–S: + |
| McPherson et al. (2014) | Family and community SC. | Indicators of family SC: Family structure, quality of parent-child relations, adult interest in the child, parent's monitoring of the child, extended family support and exchange. social support networks Indicators of community SC: civic engagement in local institutions, trust and safety, religiosity, quality of the school, quality of the neighborhood. Other indicators were classified as “other family or community SC” or “composite SC”. | Some precise information | Family SC (most consistent evidence for parent-child relationships and extended family relationships) and community SC (most consistent evidence for both children's and parent's quality and quantity of social support networks, quality of school and neighborhood) is related to better mental health and fewer behavioral problems. However, there were also a number of non-significant and negative relationships. There is evidence that this association is stronger in affluent and low-violence neighborhoods and for children at schools in urban compared to rural communities. | I-mixed: +, n.s., - E-mixed: +, n.s., - |
| McPherson et al. (2013) | Family and community SC | Indicators of family SC: Family structure, quality of parent-child relations, adult interest in the child, parent's monitoring of the child, extended family support and exchange. social support networks Indicators of community SC: civic engagement in local institutions, trust and safety, religiosity, quality of the school, quality of the neighborhood. Other indicators were classified as “other family or community SC” or “composite SC”. | Some precise information | There was good evidence that family (most consistent evidence for parent-child-relations and relations with other family members) and community SC (most consistent evidence for quality and quantity of social support networks of children and their families, civic engagement, group activities, quality of school and neighborhood) is related to better mental health and fewer behavioral problems. Yet, here were also a number of non-significant and negative relationships. Variations are observed due to e.g. ineffective coping- networks or increase health-risk-behaviour in support networks, living in urban compared to rural regions, belonging to minority groups, living in one compared to two-parents-households, being female and younger versus older adolescents. | I-mixed: +, n.s., - E-mixed: +, n.s., - |
| Murayama et al. (2012) | Multi-level SC comprising community, workplace (ecological level, aggregated or contextual indices), and individual-level SC. SC dimensions or indicators are described as they are used in the reviewed studies, but the authors did not group the reviewed studies according to SC dimensions. | A broad range of indicators were assessed. Frequently used indicators for community- or workplace level SC: Trust and civic participation (aggregated variables); voter turnout (contextual indicator) |
Some precise information | Evidence from prospective multilevel studies indicates that both individual and ecological level (community and workplace) SC can have positive effects on different health outcomes, although there were also a number of non-significant relationships. | All cause-mortality: +/n.s./- Suicide: +/n.s. Alcohol-related mortality: + Cancer-related mortality: + Hospitalization: for CHD, psychosis: +, for depression: n.s. Depression: +/n.s. Self-rated-health: + |
| Nyqvist et al. (2014) | Cognitive and structural SC. |
Indicators of cognitive SC: social support, trust. Indicators of structural SC: social participation, social networks. |
Some precise information | Longitudinal studies provide evidence for a positive relationship between higher individual-level-structural SC and length of life. This was most pronounced regarding the structural SC indicator social participation, followed by social networks. There was some evidence regarding the cognitive SC indicators of trust, and no evidence regarding social support. | I–C: + I–S: ++ With moderators (age and gender): attenuated |
| Ransome et al. (2018) | SC as one whole on individual or ecological levels. | Indicators of SC: Social/civic participation, social trust, social cohesion, social support, collective efficacy, social control; or composite indices. | Some precise information | A protective effect of SC against HIV could be found in 58% of the studies, whereas there were also numerous associations pointing to a negative impact of SC on HIV or no relationships. Different indicators of SC were either associated with higher or lower diagnoses rates or HIV care, which might be explained by confounding variables such as social segregation and social conditions. | +/n.s./- |
| Uphoff et al. (2013) | Cognitive, structural, bonding, bridging, and linking SC as individual or contextual SC (area-/community-level), including multi-level. Results are further grouped according to functions that SC can have in the relationship between socioeconomic status and health. | A broad variety of indicators that are used by the reviewed studies are reported, but it is not clear how the authors grouped specific indicators into the dimensions. Example indicators: friendship, trust, religious participation. | No precise information | There was a general positive relationship between SC, socioeconomic status and health, when SC was measured on an individual level. When SC was measured on the contextual level, this relationship was non-significant. Regarding bonding SC, there was consistent (individual level) and some (ecological level) evidence that the positive relationship with health was more pronounced in low-SES- and minority groups (buffer effect). For bridging and linking SC, there was some evidence that the positive relationship with health was more pronounced in high-SES compared to low-SES and minority groups (dependency effect). Also, these SCs were negatively related to health for low-SES and minority groups, especially if the latter lived in areas with high bridging and linking contextual SC. | I: ++ E: n.s. With moderators: I–C, I–S: +/n.s./- E-C, E-S: +/n.s./- |
| Villalonga-Olives and Kawachi (2017) | SC as a broad category that includes individual, community and multiple levels. SC dimensions or indicators are described as they are used in the reviewed studies. The authors group studies according to the negative functions of SC. In the discussion they summarize results for community, bonding, and bridging SC. | A broad variety of indicators used by the reviewed studies are reported. These indicators present both specific items, scales, and constructs (e.g., trust, community participation), as well as broader dimensions of SC (e.g., cognitive, structural, bonding, bridging, linking). | Some precise information | There is evidence that high bonding SC (especially in the presence of low bridging SC) can be related to lower health outcomes. Community SC can be positively or negatively related to health depending on specific subgroups or contexts (e.g., gender, young age, negative relationship for low-trusting individuals in regions with high community SC). Mechanisms explaining the negative relationships might be related to exclusion of outsiders, excess claims on group members, restrictions on individual freedoms, downward leveling norms, social contagion and cross-level interactions between social cohesion and individual characteristics. | C: S: −/+ |
| Villalonga-Olives et al. (2018) | SC as a broad category that includes individual, community and multiple levels. SC dimensions or indicators are described as they are used in the reviewed studies. The authors did not group the reviewed studies according to SC dimensions. Instead, they focus on the level and the function of SC (as a target and/or channel, or segmentation variable in interventions). |
A broad variety of indicators used by the reviewed studies are reported. These indicators present both specific items, scales, and constructs (e.g., trust, community participation), as well as broader dimensions of SC (e.g., cognitive, structural, bonding, bridging, linking). | Some precise information | 8 out of 9 manuscripts show a positive effect on SC and/or the health outcomes evaluated after the intervention. Yet, it is often not tested whether an improvement in health is due to an improvement in SC. | + |
Notes: When only indicators of SC were presented in the results (e.g., trust), we re-classified these into “cognitive” or “structural” dimensions at either the individual or ecological level in order to more easily compare results across reviews.I–C (individual level cognitive); I–S (individual structural); I-mixed (individual mixed SC); E-C (ecological cognitive); E-S (ecological structural); E-mixed (ecological mixed SC); no indication = mixed SC.
++ → strong evidence that SC is associated with better health/less illness; + → some evidence that SC is associated with better health/less illness; - → evidence that SC is associated with worse health outcomes/more illness; n.s. → non-significant findings.