Findings from a Peer-Facilitated, Social Isolation Intervention in the Veterans Health Administration Healthcare System: A Mixed-Methods, Pilot Feasibility Study

Abstract

Background

Social isolation is a global public health threat. Veterans are particularly at risk for social isolation due to high rates of comorbid physical and mental health problems. Yet, effective interventions are limited.

Objectives

Our primary objective was to assess the feasibility and acceptability of CONNECTED, a novel, transdiagnostic intervention to reduce social isolation that includes individual and group components and is delivered by peers via telehealth. Secondary objectives were to identify appropriate outcome measures and explore preliminary intervention effects.

Methods

This was a two-phase study. In Phase 1, to evaluate study feasibility, we surveyed 200 veterans to assess prevalence of social isolation and their interest in social connectedness interventions. In Phase 2, we employed a mixed-methods, pre-post study design in which we piloted CONNECTED with 19 veterans through 2 successive cohorts to further assess feasibility, to evaluate acceptability, and to explore preliminary effectiveness. Quantitative analyses involved descriptive and bivariate analyses as well as multivariate modeling. Qualitative interviews were analyzed using thematic analysis.

Results

For Phase 1, 39% of veterans surveyed were socially isolated. Participants who were ≤ 55 years old, caregivers, and those who experienced unmet social needs were more likely to report social isolation. Over 61% expressed interest in VA programs to reduce social isolation. For Phase 2, the pilot intervention, recruitment rate was 88% and the enrollment rate was 86%. Retention rates for the two cohorts were 80% and 50%, respectively, and satisfaction rates among intervention completers were 100%. Results also showed statistically significant improvements in social isolation (+ 5.91, SD = 4.99; p = .0028), social support (+ 0.74, SD = 1.09; p = .03), anxiety (-3.92, SD = 3.73; p = .003), and depression (-3.83, SD = 3.13; p = .001). Results for the other measures were not statistically significant.

Conclusion

CONNECTED is a feasible and acceptable intervention and is likely to be an effective tool to intervene on social isolation among veterans.

Supplementary Information:

The online version contains supplementary material available at 10.1007/s11606-023-08387-x.

BACKGROUND AND OBJECTIVES

Social isolation is a critical public health concern and a major risk factor for poor health^1–7. Unlike loneliness, which refers to perceptions of social isolation and can occur in the presence or absence of social isolation, social isolation is an objective state characterized by the absence or paucity of social connections^{8, 9}. Prior to the COVID-19 pandemic, public health concerns about social isolation were growing globally¹⁰. Moreover, since the COVID-19 outbreak, rates of social isolation in the U.S. have increased^{11, 12}. Furthermore, the negative impacts of social isolation on health are substantive and well-documented. Adults who are socially isolated report worse physical health¹³, have more physician visits¹⁴, and experience higher rates of mortality than those who are not.¹ Social isolation also contributes to cardiovascular diseases, dementia, depression, suicidal ideation, and premature death^{14, 15}.

Compared to the general population, veterans are more vulnerable to factors that contribute to social isolation, placing them at higher risk. They are more likely to be divorced, to live alone^{16, 17}, to be homeless ¹⁸, and to have a mental illness¹⁹. Among veterans, social isolation is associated with depression, low social functioning, and suicidality^{20, 21}. Socially isolated veterans are up to 5 times more likely to be hospitalized for mental health conditions compared to those with low-levels of social isolation²². Conversely, social connections are associated with lower depression and fewer mental health visits among veterans ^{20, 23}. Data about the prevalence of social isolation among veterans are scarce, and existing studies are limited to specific subgroups of veterans²⁴. One study of older veterans (65 years of age or older) who were Medicare beneficiaries estimated that 1 in 5 were socially isolated ²⁴. In 2022, the Annual Warrior Project reported that 78% of veterans surveyed were socially isolated, a sharp increase from the 2019 rate of 43%^{25, 26}.

Given the well-documented deleterious effects of social isolation, effective strategies to mitigate these effects are needed. Indeed, social isolation has been identified as a modifiable negative social determinant of health, (SDH)^27–29. Despite this, most patients are not asked about social isolation in healthcare settings, even though patients have expressed openness to screening and desire for assistance with social isolation³⁰. Although several social isolation interventions have been developed, robust data on their effectiveness are limited³¹.Additionally, most social isolation reduction programs were developed with samples that lack racial, age-, and gender-diversity^{15, 27–29}. To address this gap, we developed the Increasing Veterans’ Social Engagement and Connectedness (CONNECTED) program—a transdiagnostic intervention aimed at reducing social isolation for patients in the Veterans Health Administration (VHA) healthcare system.

RESEARCH DESIGN AND METHODS

This was a two-phase study conducted from February to September 2022 in primary care clinics at a large Midwestern Veterans Affairs Medical Center. In phase 1, study feasibility, we surveyed patients to estimate: a) prevalence of social isolation, and b) interest in participating in an intervention to address social isolation. In Phase 2, we piloted the intervention to evaluate its feasibility and acceptability using a mixed-methods approach. All participants provided verbal informed consent prior to participation. The study was approved by the local university Institutional Review Board and the Veterans Affairs Medical Center. It followed the standard CONSORT extension guidelines for pilot feasibility trials³², which was adapted for this non-randomized feasibility study, and the Template Intervention Description and Replication Checklist (Appendix 1).

Phase 1: Survey to Establish Study Feasibility

Study Setting, Sample, and Recruitment Procedures

In phase 1, we sought to a) recruit a purposive, non-representative sample of 200 participants to evaluate the feasibility of our recruitment methods and b) identify participants for the Phase 2 pilot intervention. We generated a list of potential participants using Veterans Health Administration (VHA) administrative data. Eligibility criteria included completion of a primary care visit at the medical center within 6 months prior to study enrollment and availability of an email address. We emailed potential participants, followed by a phone call, inviting them to complete a brief, 15-min survey over the phone with a research assistant about their experiences with social isolation. To ensure that we captured the views of female and minoritized veterans, we oversampled these groups. We contacted 50 Veterans at a time, starting with those who had the most recent primary care visits, and continued recruitment until we had 200 surveys completed. Participants were offered $5 gift card upon survey completion and a chance to win an additional $50 gift card in a lottery.

Measures

Social isolation was measured with the Lubben Social Isolation Network Scale (LSNS-6). The LSNS-6 is a validated, 6-item, self-reported measure that evaluates the frequency, size, and closeness of contact for respondents’ social networks. Items are scored on a 6-point Likert scale, ranging from 0 to 5, with higher values representing more robust social networks. Scores range from 0 to 30, with ≤ 12 indicating social isolation³³. The scale has been used with a variety of ages, racial groups, and settings^34–36. The LSNS-6 has strong internal consistency (α = 0.83)³³. Other measures included questions about participants’ views of social isolation, experiences of unmet social needs (food insecurity, housing instability, and lack of transportation), and interest in social isolation reduction interventions. (Survey is included in Appendix 2).

Phase 2: Intervention Pilot

Sample and Recruitment Procedures

The objective for Phase 2 was twofold: a) to evaluate feasibility and acceptability of the CONNECTED intervention, and b) to examine potential, preliminary effectiveness of CONNECTED to inform the design of an effectiveness trial. We supplemented the quantitative data with qualitative interviews about participants’ experiences with the intervention and its perceived effects.

To be eligible, participants must a) have reported social isolation (i.e., a score of ≤ 12 on the LSNS-6) in the Phase 1 survey, b) expressed interest in participating in social isolation reduction interventions, and c) be available to meet during group sessions. Participants completed outcome assessments at baseline and 3 months and received $25 for the baseline and $50 for the 3-month assessment. Assessments took average 30 min to complete. Figure 1 presents the timeline of data collection and intervention activities.

Figure 1.

CONNECTED Study timeline of data collection and intervention activities.

CONNECTED Intervention

Grounded in Wang and colleagues’ social isolation conceptual model³⁷, CONNECTED is a manualized intervention delivered by VHA peer support specialists (peers) via telehealth through weekly individual and group sessions over 3 months. CONNECTED has the following components: a) peer support, an evidence-based approach that includes person-centered assessment of factors driving social isolation to foster patient self-efficacy³⁸; b) psychosocial interventions to address social isolation; and c) navigation, a well-established care model that addresses SDH by connecting individuals to resources that meet their health and psychosocial needs³⁹ to expand their social networks and increase social engagement. Through these intervention components, we aimed to improve the quantity (e.g., social network size), structure (e.g., diverse sources of social connections), function (e.g., perceived support), and quality (e.g., positive social bonds) of participants’ social relationships.

CONNECTED consisted of 2.5 peer contact hours per week, for a total of 30 h. Group sessions lasted 90-min and were delivered via VA TEAMS. Individual sessions lasted 30–60 min and occurred through TEAMS or phone, based on participants’ preferences. Individual sessions sought to engage patients, address barriers to group participation, create opportunities for socialization, and facilitate self-efficacy for social engagement and achievement of social connectedness goals. Group sessions drew from elements of cognitive behavioral therapy, motivational interviewing, and acceptance and commitment therapy to target psychosocial barriers to social engagement and support development of skills to increase social connectedness. Sessions incorporated didactics, hands-on activities, small-group discussions, and opportunities for social engagement. An overview of group content is included in Supplement Table 1. CONNECTED also included a patient workbook with worksheets and educational materials that reinforced group lessons. Peers provided individualized navigation services for social resources within and outside the VHA, facilitated by collabortation with the VHA Whole Health Program and Recreational Art Therapy Services, as well as several community organizations. Two peers (one Black male, one White female), who completed over 40 h of intervention-specific training, delivered the intervention. Some peer-led social and community activities were conducted in-person as part of the navigation component of the intervention. Each participant was assigned a peer for the individual sessions based on preference (e.g., shared gender or racial identity, or schedule).

Phase 2 measures

Feasibility and Acceptability

Study feasibility was determined by the ability to recruit, enroll, and retain participants, and administer study procedures. Feasibility of the CONNECTED intervention was assessed based on peers’ delivery of the intervention with fidelity to study protocol, and participants’ treatment adherence (number of sessions completed). To assess fidelity, two clinical psychologists observed in-vivo one randomly selected group session and completed the CONNECTED fidelity scale. The scale assessed fidelity at the peer level and included evaluation of session content and peers’ clinical competence. Scores ranged from 0–20, with a score of 16 (80%) indicating fidelity. Feedback based on fidelity assessments was provided to peers to refine intervention delivery and reinforce training.

Intervention acceptability was assessed quantitatively via survey and qualitatively through interviews across three domains from the Theoretical Framework of Acceptability: affective attitude (how participants feel about the intervention), burden (challenges of the intervention), and perceived effectiveness (perceived usefulness of the intervention)^{40, 41}. We assessed satisfaction with an 5-item survey developed for this study (see Supplemental Appendix A), which evaluated satisfaction with the peers, specific components of the intervention, and the intervention overall. All items were ranked on a 5-point Likert Scale, ranging from 1 = very satisfied to 5 = very dissatisfied. Our pre-determined benchmark for acceptability was to have 80% of participants reporting satisfaction with the program. The qualitative interviews were guided in part by the health equity implementation framework⁴² and included questions about a) experiences with the intervention; b) views of the peers; and c) views of the intervention, its components, and perceived impact (See supplemental Appendix B). Interviews, which lasted 30–45 min, were conducted via video by a research assistant. They were audio-recorded and transcribed for analysis.

Preliminary Effects of the Intervention

We administered several measures to identify appropriate patient-centered outcomes for the future trial and to evaluate potential effectiveness of CONNECTED. These included measures of social isolation, loneliness, depression, anxiety, and quality of life. Social isolation was assessed with the Lubben Social Network Survey (LSNS-6), described above³³. Loneliness was measured with the UCLA Loneliness Scale -6 (ULS-6)⁴³. It is scored on a 4-point Likert Scale, ranging from 1 to 4, with lower scores indicating less loneliness. The ULS-6 and has demonstrated internal consistency (α = 0.89—0.94) and test–retest reliability (r = 0.73)⁴³. Anxiety was assessed with the General Anxiety Disorder-7 (GAD-7) scale, a validated 7-item self-report measure that has shown high internal consistency (α = 0.89), and good sensitivity. A score of 10 or higher indicates moderate to severe anxiety⁴⁴. Depression was measured with the Patient Health Questionnaire (PHQ-9), a 9-item scale with high internal consistency (α = 0.89) and good sensitivity. A score of 10 or higher indicates moderate to severe depression⁴⁵. The Multidimensional Scale of Perceived Social Support (MSPSS) was included to assess perceived adequacy of social support. The scale is measured on a 5-point Likert Scale, with a score of 3 or higher indicative of moderate to high support ⁴⁶. The Life Engagement Scale (LET) is a 6-item survey that measures the extent to which a person engages in activities that are valued; the scale has high internal consistency (α = 0.80)⁴⁷. Items are measured on a 5-point Likert Scale with higher scores indicative of higher engagement. Quality of Life was assessed with the Veteran Rand Health Survey (VR-12), a 12-item self-administered questionnaire that assesses perceptions of general physical, social, and mental health functions⁴⁸. We supplemented the quantitative data with qualitative interviews about experiences with the intervention and its perceived impacts. We also collected demographic data at baseline and conducted chart reviews to collect patients’ mental health diagnoses at the end of the study.

DATA ANALYSES

Quantitative Data

Basic descriptive statistics for patients’ characteristics were calculated as means (standard deviations) for continuous variables and frequencies (percentages) for categorical variables. Bivariate analyses were performed using Student’s t-tests and Fisher’s Exact tests (Mantel–Haenszel tests were used for ordinal/ranked data), respectively. Fisher’s Exact tests were used due to many analyses having small expected cell counts, but all such analyses were verified against the general Chi-Square test. A multivariable model was then performed to determine which of the bivariate results would remain independently associated with isolation using bivariate results that were significant at p < 0.10, after removing any necessary variable that were collinear or had missing data that decreased the sample size. All analytic assumptions were verified, and analyses were performed using SAS v9.4⁴⁹.

Qualitative Data

We used an inductive and deductive thematic analysis approach to analyze the data⁵⁰. This involves identifying and comparing common emergent themes across transcripts and using pre-identified concepts from our research questions to organize and analyze the data. A 4-member analytical team that included the first author independently read 3 of the transcripts to gain a general understanding of the data and variations across participants. We generated a working set of codes that included a priori codes from the health equity implementation framework. Once we had a well-defined and established set of codes, two coders independently coded each transcript. Then, the team reviewed and finalized the coding to maintain consistency and consensus in our coding practice. We then collaboratively identified themes that emerged from data.

Mixed-Methods Data Integration

We used a convergent mixed-methods design to analyze the study, using both quantitative and qualitative data concurrently and iteratively to inform the results^{51, 52}.

RESULTS

Phase 1 Results: Prevalence of Social Isolation and Patient Characteristics

As shown in the CONSORT Figure, Phase 1 participants were recruited from 22,366 potentially eligible patients. We reached out to a non-representative sample of 381 veterans and successfully contacted 260. Of those, 200 completed the survey, resulting in a response rate of 80%. Survey results are presented in Table 1. Participants were mostly non-White (52%), male (73%), employed full-time (53%), married (60%), and 40% had at least some college education. Of the 200 patients surveyed, 39% met the threshold for social isolation. Among the isolated participants, their average LSNS-6 score was 8.2 (SD = 3.3, range 0–12). Participants who had children living with them (26.3% vs. 58%, p < 0.0001), identified as a caregiver (26% vs. 47%, p = 0.0088), were < 55 years old (40% vs. 58%, p = 0.0105), and reported any unmet social needs (14% vs. 26%, p = 0.03) were more likely to be socially isolated. Although not statistically significant, relative to White participants, more non-White participants were socially isolated (48% vs. 60%). Moreover, relative to non-isolated Veterans, isolated participants reported significantly worse physical (6.4 vs. 5.4; p = 0.001) and mental health (7.4 vs. 5.8; p < 0.0001). Both isolated (61%) and non-isolated (51%) participants expressed high levels of interest in participating in VA programs to reduce social isolation (Fig. 2).

Table 1.

CONNECTED Survey Results

	Overall (N = 200) Mean, (SD)	Isolated NO (n = 123) Mean, (SD)	Isolated YES (n = 77) Mean, (SD)	P-value
Social Isolation (LSNS-6 Score)	14.9 (6.7); 0 – 30	19.1 (4.6); 13 – 30	8.2 (3.3); 0 – 12	< 0.0001*
Sex
Male   Female	146 (73.0) 54 (27.0)	91 (74.0) 32 (26.0)	55 (71.4) 22 (28.6)	0.6921
Race
White   Non-White	95 (47.5) 105 (52.5)	64 (52.0) 59 (48.0)	31 (40.3) 46 (59.7)	0.1047
Ethnicity (Latino)
Yes   No	10 (5.0) 190 (95.0)	3 (2.4) 120 (97.6)	7 (9.1) 70 (90.9)	0.0471*
Age
25–34   35–44   45–54   55–64   65–74   75–84   85 +	26 (13.0) 28 (14.0) 40 (20.0) 46 (23.0) 46 (23.0) 13 (6.5) 1 (0.5)	16 (13.0) 15 (12.2) 18 (14.6) 32 (26.0) 31 (25.2) 10 (8.1) 1 (0.8)	10 (13.0) 13 (16.9) 22 (28.6) 14 (18.2) 15 (19.5) 3 (3.9) 0 (0)	0.1580
Age (comparing > 55 vs. 55 +)
18–54   55 +	94 (47.0) 106 (53.0)	49 (39.8) 74 (60.2)	45 (58.4) 32 (41.6)	0.0105*
Education
< HS-Trade   Some-Associates   Bachelor’s   Master’s—PhD	45 (22.5) 81 (40.5) 44 (22.0) 30 (15.0)	30 (24.4) 51 (41.5) 27 (22.0) 15 (12.2)	15 (19.5) 30 (39.0) 17 (22.1) 15 (19.5)	0.1691
Sexual Orientation
Hetero/Straight   Other	195 (97.5) 5 (2.5)	120 (97.6) 3 (2.4)	75 (97.4) 2 (2.6)	> 0.9999
Relationship status
Married/partner   Other	121 (60.5) 79 (39.5)	79 (64.2) 44 (35.8)	42 (54.6) 35 (45.5)	0.1729
Employment status
Full time   Part time   Retired   Unemployed   In school   On disability	106 (53.0) 10 (5.0) 60 (30.0) 9 (4.5) 2 (1.0) 13 (6.5)	64 (52.0) 6 (4.9) 41 (33.3) 3 (2.4) 1 (0.8) 8 (6.5)	42 (54.6) 4 (5.2) 19 (24.7) 6 (7.8) 1 (1.3) 5 (6.5)	0.4608
Household income
Comfortable   Just enough   NOT enough	124 (62.0) 62 (31.0) 14 (7.0)	84 (68.3) 32 (26.0) 7 (5.7)	40 (52.0) 30 (39.0) 7 (9.1)	0.0621
Lives alone
No   Yes	156 (78.0) 44 (22.0)	99 (80.5) 24 (19.5)	57 (74.0) 20 (26.0)	0.2972
Children live with you
No   Yes	97 (62.2) 59 (37.8)	73 (73.7) 26 (26.3)	24 (42.1) 33 (57.9)	0.0001*
Caregiver
No   Yes	103 (66.0) 53 (34.0)	73 (73.7) 26 (26.3)	30 (52.6) 27 (47.4)	0.0088*
Military Branch
Army   Navy   Marines   Air Force   National Guard   Other	112 (56.0) 24 (12.0) 22 (11.0) 39 (19.5) 2 (1.0) 1 (0.5)	69 (56.1) 16 (13.0) 9 (7.3) 27 (22.0) 2 (1.6) 0 (0)	43 (55.8) 8 (10.4) 13 (16.9) 12 (15.6) 0 (0) 1 (1.3)	0.1485
Era of Service
Vietnam   Korean   Gulf   OEF/OIF/OND   Other	42 (21.0) 6 (3.0) 43 (21.5) 65 (32.5) 44 (22.0)	31 (25.2) 3 (2.4) 23 (18.7) 40 (32.5) 26 (21.1)	11 (14.3) 3 (3.9) 20 (26.0) 25 (32.5) 18 (23.4)	0.3491
Concerned about not feeling connected
Not concerned   Somewhat concerned   Very concerned	127 (63.5) 49 (24.5) 24 (12.0)	90 (73.2) 24 (19.5) 9 (7.3)	37 (48.1) 25 (32.5) 15 (19.5)	0.0010*
Interest (in VA program)
No   Yes	90 (45.0) 110 (55.0)	60 (48.8) 63 (51.2)	30 (39.0) 47 (61.0)	0.1909
Run out of food
No   Yes	187 (93.5) 13 (6.5)	117 (95.1) 6 (4.9)	70 (90.9) 7 (9.1)	0.2531
Worried about stable housing
No   Yes	184 (92.0) 16 (8.0)	116 (94.3) 7 (5.7)	68 (88.3) 9 (11.7)	0.1794
Lack of transport
No   Yes	184 (92.0) 16 (8.0)	116 (94.3) 7 (5.7)	68 (88.3) 9 (11.7)	0.1794
Food/Housing/Transport ANY
No   Yes	163 (81.5) 37 (18.5)	106 (86.2) 17 (13.8)	57 (74.0) 20 (26.0)	0.0395*
Physical health (low scores → dissatisfied; high scores → satisfied); range 1–10	6.0 (2.1); 1 – 10	6.4 (1.9); 1 – 10	5.4 (2.2); 1 – 10	0.0010*
Mental health (low scores → dissatisfied; high scores → satisfied); range 1–10	6.8 (2.4); 1 – 10	7.4 (2.2); 1 – 10	5.8 (2.4); 1 – 10	< 0.0001*

Values are means (standard deviation); ranges for continuous variables and frequencies (percentages) for categorical variables, with p-values from t-tests and Fisher’s Exact tests (due to the typical small cell counts; Mantel–Haenszel Chi-Square tests were used for ordinal/ranked data), respectively

Bold numbers with asterick indicate statistical significance

Figure 2.

CONNECTED CONSORT Diagram.

In Phase 1 we also identified 63 survey completers who were socially isolated and interested in participating in the intervention (Phase 2). Of those 63, 25 were successfully contacted and deemed eligible to participate in the intervention by providing informed consent for participation and being available for group meetings.

Phase 2 Results

Twenty-two participants were recruited from Phase 1 (88%); 19 enrolled (86%) in the intervention. The first ten 10 formed cohort 1 and the other 9 formed cohort 2. Two participants withdrew from the first cohort after the first session due to schedule conflicts, resulting in 80% retention. In the second cohort, one participant withdrew before the intervention began for mental health reasons, leaving 8. We retained 50% (4/8) of these participants. Retention in this cohort was lower because one of the peers left the study team halfway through the intervention. There were no changes made to the intervention following the peer’s departure. The peer’s participants were re-assigned to another peer, whom they had met. Some participants nonetheless disengaged in the study due to difficulties forming a relationship with the new peer.

Intervention Feasibility

Peers successfully delivered the intervention to study participants with fidelity, earning “satisfactory” for most fidelity elements and a total score of 18 out of 20 (90%). In terms of session adherence, 5 of 8 participants from cohort 1 completed at least 10 of the 12 group sessions. For cohort 2, as discussed above, we retained 4 (50%) of study participants due to a peer leaving the study. Of those 4 participants, 3 completed ≥ 10 group sessions.

Intervention Acceptability

Satisfaction rates with the intervention were 100% among intervention completers across both cohorts, and 85% of participants reported being “very satisfied” with CONNECTED. Qualitative interviews with 10 intervention completers (83%) corroborate these findings. Specifically, participants expressed satisfaction with the peer-facilitation component of the intervention, the telehealth modality, and the generalist approach, which emphasized inclusion of diverse patient groups across diagnostic categories. Two participants described their experiences with the program and how it benefited them as follows:

The very first group that I had I was so energized. …. I was so happy even though I did feel like, “maybe this isn’t for me.” ... And my husband could see it on my face…[group] just made me feel extremely happy… I really looked forward to every Wednesday, and every one-on-one phone call. -C06

COVID forced us into social isolation. And [CONNECTED] kind of does the opposite. It

forces you out of the social isolation so that you make a commitment to get out once a

week and …having interactions with other people…to start living outside the house again. And I like that you didn't have to just jump out with two feet. You had a platform to stand on to help you start engaging with people again. -C114

Participants described positive effects of the intervention beyond alleviating social isolation. They noted that the intervention improved their depressive and anxiety symptoms, helped them develop new skills, and enhanced their overall quality of life.

It’s good to see that the VA is really starting to look at ... so many different ways to deal with what veterans need, because we've just been through things that a lot of people don't experience. And giving those opportunities, and not just say, here, take this pill and go home. They're going to try to help you learn, grow, thrive, and become more, a better functioning person. -C04

[CONNECTED] definitely helped. I was becoming a little bit more reclusive…So, it's been good for me in that I've started to open back up and reaching back out to people…My wife has always been more social. But she started to become more reclusive with me, and now she's becoming more social again, too. -C26

I realized how I approached things and how I thought of certain people was the wrong way…which is basically self-isolating without realizing it. So, doing this program with all the other veterans in the group, [I realized] I need to break out of this. -C141

Preliminary Effects

Nineteen veterans participated in the CONNECTED intervention. Baseline characteristics are presented in Table 2. Most were non-White (58%), male (53%), married (58%), did not live alone (79%), had at least an Associate’s degree (58%), were employed or full-time students (58%), had a mental health diagnosis (63%), and described their household income as being “comfortable” (63%).

Table 2.

CONNECTED Intervention Participants’ Characteristics (N = 19)

	Combined Cohorts (n = 19, %)	Cohort 1 (n = 10, %)	Cohort 2 (n = 9, %)
Age
25–34	3 (16%)	1 (10%)	2 (22%)
35–44	4 (21%)	2 (20%)	2 (22%)
45–54	8 (42%)	4 (40%)	4 (44%)
55–64	2 (10%)	2 (20%)	0 (0%)
65–74	1 (5%)	1 (10%)	1 (11%)
Sex
Male	10 (53%)	6 (60%)	4 (44%)
Female	9 (47%)	4 (40%)	5 (55%)
Race/Ethnicity
Black	6 (31%)	3 (30%)	3 (33%)
White	8 (42%)	4 (40%)	4 (44%)
Asian	1 (5%)	0 (0%)	1 (11%)
American Indian/Alaska Native	2 (10%)	1 (10%)	1* (11%)
Hispanic/Latino	3 (16%)	2 (20%)	1*(11%)
Marital Status
Single (never married)	6 (31%)	2 (20%)	4 (44%)
Married	11 (58%)	7 (70%)	4 (44%)
Divorced	2 (10%)	1 (10%)	1 (11%)
Education
High Diploma/ GED	1 (5%)	0 (0%)	1 (11%)
Some College	7 (37%)	4 (40%)	3 (33%)
Associate Degree/ Trade certificate	5 (26%)	2 (20%)	3 (33%)
≤ Bachelor’s Degree	6 (31%)	4 (40%)	2 (22%)
Employment Status
Full-time Employment/School	11 (58%)	8 (80%)	3 (33%)
Employed part-time	1 (5%)	1 (10%)	0 (0%)
Unemployed	2 (10%)	0 (0%)	2 (22%)
Retired	3 (16%)	1 (10%)	2 (22%)
Service Branch
Army	13 (68%)	7 (70%)	6 (67%)
Marines	3 (16%)	2 (20%)	1 (11%)
Air Force	2 (10%)	0 (0%)	2 (22%)
Navy	1 (5%)	1 (10%)	0 (0%)
Service Era
Gulf war	8 (42%)	5 (50%)	3 (33%)
OEF/OIF/OND	8 (42%)	4 (40%)	4 (44%)
Vietnam War	1 (5%)	0 (0%)	1 (11%)
Korean War	1 (5%)	1 (10%)	0 (0%)
None	1 (5%)	0 (0%)	0 (0%)
Income
Comfortable	12 (63%)	6 (60%)	6 (67%)
Do not have enough to make ends meet	2 (10%)	1 (10%)	1 (11%)
Have just enough to make ends meet	5 (26%)	3 (30%)	2 (22%)
Living Arrangement
Lives alone	4 (21%)	2 (20%)	2 (22%)
Does not live alone	15 (78%)	8 (80%)	7 (78%)
Mental Health Diagnoses*
Personality disorders   Mood disorders   Severe Mental Illness   Substance use disorders (SUD)   PTSD   No diagnosis	1 (5%) 7 (37%) 3 (16%) 3 (16%) 5 (26%) 7 (37%)	0 (0%) 4 (40%) 2 (20%) 1(10%) 3 (30%) 4(40%)	1 (11%) 3 (33%) 1 (11%) 2 (22%) 2 (22%) 3 (33%)

% refers to percentage of participants in each cohort

*Diagnoses were included if they were in a mental health visit note completed within a year prior to the end of study participation. Some participants had multiple diagnoses therefore the total percentage will be greater than 100

In Table 3, we show the combined (cohorts 1 and 2) results on study outcome measures. Overall, the intervention led to statistically significant improvements in social isolation as indicated by the change in scores from baseline to 3 months (+ 5.91, SD = 4.99; p = 0.002). There was also statistically significant improvement in scores for social support (+ 0.74, SD = 1.09; p = 0.03), anxiety (-3.92, SD = 3.73; p = 0.003), and depression (-3.83, SD = 3.13); p = 0.001). Although not statistically significant, participants also improved on life engagement, loneliness, and mental health functioning. There was no improvement in perceived physical health functioning. We summarize the study results in Fig. 3 below.

Table 3.

Effects of CONNECDTED on Study Outcomes (Combined Cohort Results n = 12)

	Baseline (pre-intervention) Mean, (SD)	Post-Intervention Mean, (SD)	Change in scores from baseline to post-intervention Mean, (SD)	p-value (for non-zero change)
Social Isolation LSNS-6	6.95 (3.41)	13.36 (5.95)	5.91 (4.99)	0.0028
Social Support MSPSS	4.33 (1.01)	5.23 (0.94)	0.74 (1.09)	0.0385
Anxiety GAD-7	10.00 (5.06)	4.67 (0.23)	-3.92 (3.73)	0.0039
Depression PHQ-9	11.68 (6.61)	5.50 (6.36)	-3.83 (3.13)	0.0014
Life Engagement LET	21.16 (5.04)	24.83 (6.19)	2.17 (3.51)	0.0559
Loneliness ULS-6	16.26 (4.34)	13.58 (4.91)	-1.92 (3.45)	0.0805
Physical functioning VR-12(physical)	37.79 (8.33)	43.41 (11.36)	0.95 (5.55)	0.7549
Mental Health functioning VR-12 (mental)	37.14 (7.35)	31.90 (8.81)	-1.70 (4.67)	0.5189

Values are means (standard deviations) for baseline and post-intervention and means (95% confidence intervals) for change, with p-values from one-sample t-tests for H0 = 0

Bold numbers with asterick indicate statistical significance

Figure 3.

Summary of CONNECTED pilot feasibility study results.

DISCUSSION

Despite its substantial negative health outcomes ^{2–4, 10}, effective interventions to reduce social isolation, especially among diverse patient populations, remain scarce ⁵³. In this study, we assessed the feasibility and acceptability of CONNECTED, a novel social isolation intervention for diverse patients in primary care settings. We also explored preliminary effectiveness of the intervention to inform the design of a future effectiveness trial.

Overall, results demonstrate that CONNECTED is feasible and acceptable. Findings indicate that 39% of survey participants were socially isolated, which is consistent with prior literature demonstrating high rates of social isolation and loneliness in veteran populations^{24, 25}. Notably, results show high rates of social isolation in veterans under the age of 55. This finding adds to growing reports of social isolation in young adulthood^{54, 55}. It also points to the need to assess and mitigate social isolation in this subgroup of patients who may be overlooked for services due to their age, given that most social isolation and loneliness interventions tend to focus on older adults^{27, 53, 56}. Younger adults may experience life changes such as relocation and care giving for young children or older adults that put them at greater risk of social isolation and negative health sequelae. Emerging studies also suggest that social isolation may be more detrimental for young adults compared to older adults^{54, 55,57}, which reinforces the need to intervene on social isolation in this age group.

Consistent with prior studies ^58–60, findings indicate that social isolation was more prevalent among those who reported any unmet social needs relative to those who did not. This is important because social isolation, as a negative social determinant of health, may exacerbate the effects of other negative social risk factors, such as food and housing insecurity, to impact population health. Our findings also extend prior literature by demonstrating that veterans desire assistance with social isolation.

To begin to address the lack of diversity in social isolation interventions, we piloted CONNECTED in a diverse sample of primary care veterans across age groups, socioeconomic status, sex, and diagnostic categories. Our findings suggest that CONNECTED is feasible in diverse patient populations with complex health and social needs. This is a major strength of the intervention, and if effective, CONNECTED could fill a major gap in patient care, given that most existing social isolation interventions have been tested primarily among participants who are female, White, and older ²⁷.

Acceptability of the intervention was high. All participants perceived the intervention to be effective and beneficial. Specifically, they identified the peer interventionists, telehealth modality, and inclusion of patients across diagnostic categories as helpful. Building on prior literature, which has shown that many social isolation interventions have had weak effects due to their narrow focus^{9, 27, 31}, we designed CONNECTED to target multiple dimensions of social isolation based on Wang and colleagues’ conceptual model ³⁷ and to incorporate multiple strategies to increase social connectedness. These include creating opportunities for social engagement, individualized navigation services, skill building, psychoeducation, and peer support. Results also indicate that this intense approach is acceptable to participants and provides a comprehensive method to improve social connectedness.

Although not powered to detect significant change, CONNECTED produced statistically significant improvement in social isolation scores from baseline to 3 months. Participants also had significantly improved scores for depression, anxiety, and perceived social support. Results on other measures improved but were not statistically significant. These findings suggest that CONNECTED has the potential to be an effective tool to address social isolation.

This study included a diverse, although a small, non-representative sample of veterans from one Medical Center. Replication and expansion of the survey with a larger veteran population are needed to produce generalizable findings. Similarly, future testing of CONNECTED, with some modifications to increase study retention, such as better peer coverage, in a fully-powered, randomized controlled trial is needed to evaluate effectiveness and readiness for implementation in VA settings.

In summary, this pilot study demonstrates that CONNECTED is a feasible and acceptable social isolation reduction intervention. CONNECTED is designed to be delivered by VHA peers, which may facilitate future implementation given that peers constitute a growing, yet affordable, workforce, compared to licensed clinicians. Use of peers as interventionists also contributes to the acceptability of interventions because, as veterans themselves, peers are relatable to VHA patients and are well-suited to address issues associated with social isolation, such as stigma^38,61,62. Moreover, the telehealth modality may contribute to increased reach of the intervention by eliminating transportation barriers. The combined individual and group format provides another model for social isolation interventions and should be tested more broadly to determine whether various levels of intervention intensity provide greater and sustained benefits.

Supplementary Information

Below is the link to the electronic supplementary material.

Author’s Contributions:

We would like to thank the study participants, and Mr. James Miller who contributed to the study as a volunteer peer. We are thankful for our clinical and community partners who facilitated the delivery of CONNECTED, and the research staff members, Emily Austin and Denise S. Zou, for their contributions to the study.

Funding

This study was funded by VA HSR&D SWIFT award and a VA HSR&D Career Development Award -2 16–153 to Dr. Eliacin. It also received support from the Regenstrief Institute, Inc.

Data Availability:

The data sets generated and analyzed during this study will be available from the corresponding author on reasonable request.

Declarations:

Conflict of Interest:

The authors report no conflict of interest.