Loneliness/Isolation intervention as network-building interventions

Interventions to address isolation/loneliness generally aim to improve the quantity and quality of social relations with existing or new support individuals and groups. Network-building interventions are deliberate efforts to change social networks of individuals [15]. Social network analysis (SNA) is an important approach to assess how interventions addressing loneliness/isolation change social networks. SNA is a well-established approach that focuses on the relational patterns between network actors [16, 17] rather than considering them as separate units. In other words, SNA captures the interdependencies among network actors, whereas conventional research methodologies assume independence among network participants [18].

A popular approach to study the network outcomes of loneliness/isolation interventions is assessing the change in the size of personal networks [19], or individuals’ evaluation of support networks [20, 21]. Many existing social network tools, including Berkman Syme Social Network Index [22] and Lubben Social Network Scale [23], ask about the number of individuals in each important social role (family, friends, colleagues, etc.), and usually provide a single score. Some researchers ask about the number of important individuals in respondents’ social networks [24], create numerical scores based on a mix of the number and quality of social relations [25], or ask respondent’s evaluation of the number of people who can help each other in the community [26].

In the following section, we argue that the number of ties does not fully capture the relational complexity of social networks in which one is embedded.

Network structure and well-being

Several aspects of network structure influence individuals’ perception of loneliness and support. Here we provided a few examples:

Centrality represents the prominence of an actor in the network. The simplest measure of centrality, degree centrality, is the number of network connections an individual has (e.g. the number of friends, or family members) [27]. Degree centrality is one of the most commonly measured indicators of network structure, although authors do not necessarily refer to a count of connections as a measure of centrality [19]. In the case of social support, degree centrality could either be indegree (the number of people who offer you social support or who name you as a friend) or outdegree (the number of people to whom you offer social support or whom you name as a friend). Indegree centrality is positively correlated with emotional support [28]. However, a variety of other measures of centrality are associated with well-being. For example, betweenness centrality, the extent to which an individual is on paths between other members of the network, and closeness centrality, the extent to which short paths connect the individual and other members of the network, are positively correlated with measures of wellness [29]. Indirect connections also matter. Being connected to more people indirectly (through others) decreases the likelihood of depression [30].

The density (connectedness) of networks affects well-being. Compared to degree centrality which is about the number of relations from or towards ego (respondent), density is about the overall connectedness of personal network, which should include relationships among network members not including ego. Density of personal networks was shown to predict loneliness in college students, particularly in men [31, 32]. However, some studies failed to show the association between density and subjective loneliness [33, 34], or proposed that the effect of density of personal networks on life satisfaction is moderated by an individual’s personality (as individuals may differ in terms of satisfaction by embedding in denser networks) [35].

Reciprocity (bi-directionality) of social relations matters for several reasons. Qualitative studies have found that older individuals have an easier time accepting social support when it is offered as part of a reciprocal relationship, that is, when they feel that they have given (in the case of their children) or are giving help to the person who is helping them [36, 37]. Individuals who perceive reciprocity in their relationship with their best friends feel less lonely [38]. There is experimental evidence that reciprocity is a key factor in building trust in networks [39]. Reciprocal imbalance in relationships (over-benefiting and under-benefiting) may lead to mental distress and less satisfaction with relations [40]. All of these findings suggest that reciprocity may be one key to maintaining relationships once they are established.

Network clustering, or the degree to which groups of three individuals connect completely with each other, is another important feature. In a longitudinal analysis of a large population-based study, Cacioppo and colleagues (2009) found that loneliness occurs in clusters, particularly clusters of individuals that are peripheral to the social network [41]. People in more clustered networks tend to be healthier [29]. Experiments find that the search for cooperative partners produces clustering, suggesting that clustered networks play a role in maintaining cooperation in groups [42, 43]. Of course, the influence that clustered networks have on their members has a downside. DiFonzo et al (2014) find that clustering increases the social influence of the group but also increases the tendency of groups to strongly differentiate themselves from others [44].

Based on the abovementioned evidence, we argue that interventions addressing loneliness/social isolation may affect the structure of social networks beyond merely increasing the number and frequency of social relations to ego. They may change the reciprocity of relations, may affect formation of denser clusters, or may even selectively affect some regions of one’s social network (e.g. only improving the quality of intimate relations, or leading to bridging ties to new clusters). Given the need for more in depth analysis of the structural targets of interventions for loneliness/isolation, we conducted a systematic scoping review of studies assessing the effect of interventions to address loneliness and social isolation on the structure of social networks. We aimed to map the types and components of these interventions and methods/metrics of assessing structural changes in social networks. We evaluated how these interventions can be adapted to promote connectedness in the context of societies after COVID19.

Literature search

Our search strategy involved forward/backward citation tracking of three seminal network intervention publications, Valente (2012) [15], Valente et al. (2015) [46], and Latkin & Knowlton (2015) [47], complemented by a bibliographic search in Web of Science and SCOPUS, performed in October 2019 (search strategy in S1 Appendix). To capture articles about relevant interventions for social isolation and loneliness that might not have been described using network intervention keywords, we also conducted a hand search of seminal reviews on the topic [19–21, 48]. Articles identified through this search strategy were imported into the web-based review program, Covidence [49], before undergoing title/abstract review.

Study selection

To establish consistency in the study selection and data charting process, all authors completed a trial screening process in pairs (three sets of pairs total). Authors met biweekly from March 2020 –July 2020 to discuss points of disagreement and achieve consensus pertaining to title/abstract and full-text study inclusion. We included studies that intentionally aimed to change aspects of social networks to address isolation/loneliness, measured the network structure as study outcomes and/or used them to inform interventional strategies, and were available in English. We excluded studies that only reported the number of individuals one is connected to (network size) or only provided aggregate measures of quality and quantity of relations (such as Berkman Syme Social Network Index and Lubben Social Network Scale).

Data extraction and synthesis

A data-charting form was jointly developed by the reviewers to determine variables to extract. Pairs of reviewers independently extracted a calibration set of studies. Once the team demonstrated consensus and consistency in data charting on the calibration set, the information in each included study was extracted by one author. The group regularly met to discuss and resolve uncertainties and potential disagreements, and continuously updated the data-charting form. Categories included on the data charting form were as follows: author/year, sample, study setting, design, characteristics of network actors, the nature of network relations, characteristics of the intervention(s), theoretical underpinnings of the intervention(s) according to the authors, measures of network change and other outcomes, effectiveness of the intervention according to the authors, and implications. Result syntheses consisted of three steps: 1) descriptive summary describing details about study design, 2) thematic analysis for the categorization of network interventions, and 3) consideration for how these interventions could be adapted to contexts with limited physical and in-person contact opportunities, as manifested by COVID19 restrictions. We did not conduct quality appraisals of included studies, which is consistent with PRISMA-ScR guidelines given the exploratory nature of scoping review methodology and the heterogeneity of studies included in scoping reviews.

Literature search and selection process

Of our initial 3390 references, we assessed the full texts of 233 articles, of which 17 studies were about the effect of interventions on social networks to address loneliness and isolation, of which 9 studies were excluded, as they either only reported the number of individuals one is connected to or developed single aggregate measures for quality and quantity of relations without providing information about the composition of social network and/or detailed information about relations to particular social groups/roles. We included eight studies that measured network structure before and after the intervention and/or incorporated network analysis as a component of the intervention (Fig 1). Three studies assessed interventions focusing on individual-level network building, and four studies assessed group-level network building interventions. Kasari et al. (2016) compared an individual-level with a group-level intervention, and consequently was included in both sections [50].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. PRISMA flow diagram of the study selection.

https://doi.org/10.1371/journal.pone.0253734.g001

Individual-level network interventions

The interventions in this category generally aimed to help the individuals with limited social connections (e.g. older adults and children and adults with intellectual, mental, and physical disabilities) strengthen/activate existing social relations or engage in activities to facilitate new relations or strengthen current ties (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Study characteristics, elements of network-building interventions, and main outcome measures.

https://doi.org/10.1371/journal.pone.0253734.t001

Group-level network interventions

Interventions in this category were administered in small groups and, in addition to network-building at individual levels, also aimed to facilitate communication, support, and role-modeling among peers within the groups (Table 1).

Networks as intervention components

In this review, we identified key elements from both individual-level and group-level network-building interventions that address social isolation and loneliness in a framework shown in Fig 2. These elements, described below, consist of the following: (a) creating opportunities to build networks; (b) building social skills; (c) assessing network diagnostics; and (d) promoting network engagement.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Components of individual- and group-level network-building interventions; opportunities and barriers for post-COVID19 adaptation.

https://doi.org/10.1371/journal.pone.0253734.g002

Networks as outcomes

Very few studies of interventions to address loneliness/isolation measure the structure of social networks. Even though included studies collected data on the structure of networks, their reported outcomes were mostly limited to basic measures, such as network size, or relative centrality of actors in the network. We did not find any study that assessed if and how interventions would affect the density, clustering, and reciprocity of relations, and whether their effect would focus on certain types of relations, or certain social circles (e.g. family, or intimate relations). Much of the literature assumes that the goal of intervention is to form links with individuals, who may then offer increased social support. This effectively (and sometimes explicitly) draws on a paradigm in which an individual’s direct connections to friends and family bring various kinds of social support, most often emotional or instrumental [60]. This is somewhat disappointing, because social network structure matters beyond the simple number of connections. There is evidence that network reciprocity encourages information diffusion [61]; and individuals find clustered networks, those with multiple closed triads, to be more supportive [62]. Clustered social networks, which include closed triads in which three people all know each other, are more supportive and more powerfully influence individuals than networks with more open triads [62, 63]. The quality of social network connections is also of concern [64]. For example, it seems that network structure and quality could be easily addressed in a program such as Genie [51], and based on the authors’ description of the program, it appears to be quite feasible to implement through online audio/video platforms such as Skype or Zoom.

Additionally, studies have not addressed the time over which network changes following intervention can be expected to endure. Interventions that strengthen relationships within existing subgroups, and provide support for their sustainment might be especially promising for reducing isolation.

Implications for post-COVID19 social relations

COVID19 pandemics and its lock-down and “social distancing” consequences worsened, and yet raised attention to the widespread problem of loneliness and its impacts on mental and physical health, particularly in older adults and vulnerable populations [65–69]. To fight against the spread of COVID-19 in the communities, countries worldwide took measures, such as social distancing, stay-at-home, shelter-in-place orders, and restricting visits to residential facilities (e.g., nursing homes) [70]. The “COVID-19 connectivity paradox” posited that factors improving social connectedness seem to increase the risk of COVID-19 exposure among populations vulnerable to the negative impact of the pandemic as well social isolation [71]. Even with the new hopes invoked by the widespread administration of vaccines, the likelihood of new mutations and upcoming seasonal waves will limit many natural opportunities for face-to-face social communication and network-building for the time being.

Network-building interventions to mitigate loneliness in individuals most in need should consequently be adapted to the new context, and to balance the need for safety and connectedness [71]. Many of these key elements could be implemented online and the trainings and preparations do not seem to be heavily burdensome for health care systems. In this scoping review, we developed a conceptual framework for various potential components of network-building interventions (Fig 2), and discussed various strategies applied in interventional studies to operationalize them. Even though this review does not synthesize the evidence of effectiveness of intervention components and potential synergistic effects of their combinations, we argue that our framework can provide guidance for intervention developers to decisively choose and mix intervention components to develop potentially effective interventions that work, even if some components are not as feasible or as effective as pre-COVID19 conditions. Our review can provide practical insights into the design and implementation of network-building interventions and potential considerations in adapting them to the needs of different target groups. In Fig 2, we summarized potential opportunities and barriers to the implementation of each key element of network-building interventions.

Since the early stages of the COVID19 pandemic and in subsequent months, many health care systems adopted various telehealth care models [72, 73], so individuals connected to institutional or community services still, at least partially, benefit from their ties with professionals and staff virtually. In other words, social workers, home health aides, and other professionals connected to populations at risk of social isolation and loneliness may be well-positioned to conduct network interventions. Their assistance is particularly valuable to people without natural connections within a household (e.g. widowed older adults). For instance, staff and volunteers in programs such as Meals on Wheels serving homebound older adults, are an obvious place to start. Consumers of such formal services could then suggest other peers who might benefit, leading to a snowball sample. Community-based programs have also been found to enhance the neighborhood social networks of older adults, potentially mitigating the risk for social isolation and loneliness [26]. Another possibility to expand social networks is to leverage the skills of paraprofessionals or even volunteers, rather than clinicians, as is done in self-help groups. While the successful Network Support program uses clinicians to facilitate the integration of recovering alcoholics into networks of sober peers [74], twelve-step programs have accomplished the task for decades with and without professional assistance [75]. There are also programs that teach individuals’ trusted natural ties, family and friends, to deliver social network interventions [25, 55]. The phrase, “Each one teach one,” could easily be modified to, “Each one reach one.”

Although virtual interactions may not completely replace the benefits of in-person contacts, studies demonstrated that older adults who maintained a similar level of virtual social interaction (e.g., social media, phone) with people outside of the household had lower levels of depressive symptoms than those who increased or reduced their virtual interactions during the COVID-19 pandemic [76]. One plausible explanation is that maintaining similar levels of virtual connections allows individuals to receive support without feeling overwhelmed by the stress and worry about the pandemic, which can also be delivered through social networks [76]. Since structural, social, interpersonal, and intrapersonal factors contribute to social isolation and loneliness [59], different components of network-building interventions (e.g. mapping, self-reflection, and creating networking opportunities) might be helpful within different contexts. Mapping, reflection, and maintenance of existing networks could be more helpful when existing ties have untapped potential.

Scalability is an issue with a number of the social network interventions identified in our review. Not surprisingly, a number of the interventions were developed and tested using clinicians, who may be prohibitively expensive during a large-scale crisis. If scale is a challenge for interventions that focus on helping individuals to map and understand their networks, interventions that create groups for isolated individuals can forge connections without needing a facilitator. However, these interventions might be limited during social distancing. One possibility is to create groups online based on common interests. It might also be possible to construct micro-groups of three or four who would meet face to face either regularly or occasionally while practicing safe social distancing [56].