Effects of social network characteristics on mental health outcomes among United States Army Reserve and National Guard soldiers

Abstract

We sought to examine the relative salience of multiple social network structural characteristics (e.g., size, composition, quality, substance use) for understanding soldiers’ mental health symptoms (anger, anxiety, depression, PTSD). Data are drawn from soldiers (N=421) participating in the Operation: SAFETY study. Negative binomial regression models examined the relationship between ten social network characteristics and mental health outcomes, controlling for age, sex, years of military service, and deployment history. Greater number of close network ties was associated with fewer symptoms of anger, anxiety, and depression (ps <.05), but not PTSD. Having more illicit drug-using network ties was associated with greater severity of anxiety symptoms (p <.05). Finally, more days spent drinking with network members was related to higher levels of anger (p <.05). Interpersonal relationships that entail substance use are associated with greater anxiety and anger while a greater number of close ties is associated with fewer anger, anxiety, and depression symptoms.

Introduction

The impacts of an individuals’ social relationships on their own health have been well documented over the years, reflecting both direct and indirect effects. Synthesized evidence across studies has demonstrated that strong social relationships are related to longer life (Holt-Lunstad et al., 2010) and has sparked calls for the inclusion of lack of social connection as a risk factor for mortality (Holt-Lunstad et al., 2010) and a public health and policy priority (Holt-Lunstad et al., 2017; Umberson & Karas Montez, 2010). This literature on social relationships is complex, using a variety of terminology and methods. Broadly speaking, analysis of individuals’ social relationships falls into two main categories: 1) individuals’ perceptions of social support (e.g., Rueger et al., 2016; Wang et al., 2021), and 2) structural characteristics of individuals’ social networks (e.g., Homish & Leonard, 2008; Platt et al., 2014). Structural characteristics commonly include, but are not limited to, size (e.g. number of people in the network), diversity (e.g., the number of roles, social groups, or categories a person belongs to) (Campbell et al., 2021; Sripada et al., 2015), and behaviors (e.g. substance use) among people in the network (Homish & Leonard, 2008; Leonard & Homish, 2008).

There is a large body of evidence indicating the importance of social relationships specifically for mental health (Kawachi & Berkman, 2001; Rueger et al., 2016; Thoits, 2011; Wang et al., 2021), which has considered both social support and social network structural characteristics. For example, meta-analyses have demonstrated consistent and enduring relationships between social support and PTSD that are bi-directional and support both main effects and stress buffering mechanisms, such that lower social support and risk for PTSD are consistently related (Wang et al., 2021). Similar patterns have been demonstrated for the relationships between social support and depression among children and adolescents (Rueger et al., 2016). Examination of structural characteristics has demonstrated that social networks with less diversity in social roles and domains are strongly related to PTSD risk among trauma-exposed individuals, regardless of levels of perceived social support (Platt et al., 2014). This study concluded that while both dimensions are important, perceived social support may only act as a buffering factor in the context of diverse social roles in the network (Platt et al., 2014). In a study of college students, peer closeness was related to reduced anxiety and depression symptoms, while the presence of substance use in the peer network increased the odds of hazardous alcohol, tobacco, and marijuana use (Mason et al., 2014).

Mental health is of particular concern in the military context, and military service members carry a significant mental health burden (Armed Forces Health Surveillance Branch, 2021). Research on social relationships and mental health in the military has primarily focused on perceived social support within the military community (e.g., within the unit, among military service members or veterans, from the family) (e.g., Cederbaum et al., 2017; Goetter et al., 2020; Goldmann et al., 2012; Rugo et al., 2020; Vest et al., 2017), with mixed findings. For example, among U.S. National Guard soldiers, one study found that unit cohesion and social support were related to less depression and suicidal ideation (Rugo et al., 2020), while others have found no relationship between unit support and depression (Vest et al., 2017). Most consistently, research has demonstrated that lower perceived social support is associated with greater risk for PTSD (Goldmann et al., 2012; Polusny et al., 2011).

The unique characteristics of military culture and experiences, social relationships, and network composition suggests that the structural characteristics of social networks may be different for military service members than they are for civilians. For example, research indicates that veterans feel more comfortable with other veterans and that the social relationships engendered during military service may be more significant than other relationships, especially after deployment when social relationships at home may have changed or deteriorated (Ahern et al., 2015; Hinojosa & Hinojosa, 2011). However, literature on social relationships and mental health among military service members has paid much less attention to the structural characteristics of service members’ social networks, or to the broader social contexts beyond the military within which service members are embedded. Recently, a limited number of studies have begun to examine these structural characteristics of military service members’ social networks. For example, Campbell and colleagues recently examined structural network characteristics (i.e., network size and diversity) as well as perceived social support among veterans and civilians (Campbell et al., 2021). Controlling for mental health and other demographic characteristics, they found that veterans were more likely to have less diversity (defined as the # of different social groups regularly engaged) in their social networks compared to civilians, and that female veterans in particular had smaller networks and less perceived social support compared to their civilian peers (Campbell et al., 2021), providing evidence that social networks among veterans may differ from civilians. Hatch and colleagues (Hatch et al., 2013) specifically examined the relationships between common mental disorders (CMD; an aggregate of anxiety and depression symptoms) and characteristics of the social network among UK military service members, including the size of the social network, social participation, and whether or not most of the members in one’s social network were also in the military. They found that the probability of having anxiety and/or depression symptoms decreased as the size of the social network increased. The proportion of military members in the network was mostly non-significant, except for former soldiers, among whom having a mostly military-affiliated network was related to increased odds of having a CMD and PTSD (Hatch et al., 2013).

Reserve and National Guard (R/NG) soldiers may be deserving of particular attention in the examination of social network structures, due to their unique dual belonging in both civilian and military networks (Griffith, 2009). When not deployed or activated, R/NG soldiers spend the majority of their time in the civilian world, with civilian employment, and often live far from military bases or other unit members; this means that at various points in their lives, their military and civilian networks may be more or less salient (Griffith, 2009). Overlapping military and civilian social networks may have differential impacts on mental health among this population. A large study of returning U.S. National Guard soldiers, surveyed 12 months post-deployment, examined social network structural characteristics, including the size and diversity of the network (defined as the # of categories of relationship types), as well as perceived social support and unit military support (Sripada et al., 2015). In this analysis, greater perceived social and unit support were associated with lower odds of having a probable mental health condition, while network size and diversity were not (Sripada et al., 2015). Other work has provided preliminary insight into both risk and protective aspects of social network characteristics for substance use among R/NG members. Specifically, more drinking buddies and heavy drinking ties in R/NG soldiers’ social networks were related to greater risk of alcohol problems, but for deployed soldiers having a greater proportion of other military members in the network was protective against alcohol problems (Anderson Goodell et al., 2020). However, this study focused more narrowly on problematic alcohol use, and did not examine the potentially broader impact of social network characteristics on mental health.

This research aims to fill a few key gaps in the existing literature. First, this study seeks to add to the limited literature examining relationships between structural characteristics of soldiers’ social networks (rather than perceptions of social support) and mental health outcomes. Second, this study examines these relationships specifically among Reserve and National Guard soldiers, who may have unique network structures, compared to active duty members. Therefore, the goal of the current inquiry was to examine which structural social network characteristics [e.g., overall size of network, composition of the network (family, female, military affiliated), quality of the network (closeness, time spent together), substance use in the network (illicit drugs, heavy drinking, drinking buddies, time spent drinking together)] are most strongly associated with reserve soldiers’ mental health outcomes (e.g., anger, anxiety, depression, and PTSD). The second goal was to determine if these relationships differ for male and female soldiers, based on research demonstrating differences in social network structure and characteristics between men and women (e.g., Campbell et al., 2021; Homish & Leonard, 2008) and in the relationship between social support and mental health outcomes (e.g., Kawachi & Berkman, 2001; Wang et al., 2021).

Methods

Participants and procedure

Data are from the baseline assessment of Operation: SAFETY (Soldiers And Families Excelling Through the Years)], an ongoing longitudinal study, that uses a socio-ecological framework to examine individual, relationship, community, and societal level factors that impact the health and wellbeing of United States Army R/NG soldiers and their partners (Anderson Goodell et al., 2018; Heavey et al., 2017; Hoopsick et al., 2017; Vest et al., 2017). The University at Buffalo IRB, Army Human Research Protections Office, Office of the Chief – Army Reserve, and the Adjutant General of the National Guard approved the study protocol. The authors have no known conflicts of interest to report.

The Operation: SAFETY study recruited participants over a 15-month period (Summer 2014 to Fall 2015) from 47 U.S. Army R/NG units in New York State. During drill weekends, the study recruitment team provided 10-minute study overviews to explain project goals and confidentiality procedures. Soldiers were then invited to complete a one-page screening form to assess study eligibility, which was based on the following: (a) the soldier and his or her partner are married or living as married; (b) one member of the couple is a current U.S. Army R/NG soldier; (c) the soldier is between the ages of 18 and 45; (d) both partners have had at least one alcoholic beverage in the past year; (e) both partners are able to speak and understand English; and (f) both partners are willing and able to participate. A total of 411 couples, comprised of 441 soldier partners and 381 civilian partners, completed the baseline survey.

Study sample

The current analytic sample consisted of 421 current soldiers (353 males and 68 females). Soldiers were an average of 31 years old (Standard deviation (SD): 6.5), and 18.5% identified as other than Non-Hispanic White (Table 1). Approximately 30% had completed college education. Average time spent in the military was over nine years, and 60% had been deployed at least once.

Table 1.

Participant Demographics and Mental Health and Social Network Characteristics

	Total (n=421)	Males (n=353)	Females (n=68)	p
Soldier characteristics, m (SD) or % (n)
Age	31.4 (6.5)	31.7 (6.6)	29.9 (5.5)	*
Race/Ethnicity
White, Non-Hispanic	79.6% (335)	79.3% (280)	80.9% (55)
Black, Non-Hispanic	5.5% (23)	6.0% (21)	2.9% (2)
Hispanic, any race	8.3% (35)	8.8% (31)	5.9% (4)
Othera	4.8% (20)	4.3% (15)	7.4% (5)	ns
Education levelb				*
High school or some college	69.8% (294)	72.0% (254)	58.8% (40)
College completion	30.2% (127)	28.1% (99)	41.2% (28)
Years served	9.5 (6.0)	9.9 (6.2)	7.8 (5.0)	**
Deployment status (yes)	60.6% (255)	64.9% (229)	38.2% (26)	***
Mental health symptoms, m (SD)
Anger	17.9 (6.7)	17.7 (6.7)	19.0 (6.6)	ns
Anxiety	4.5 (5.5)	4.2 (5.3)	6.1 (6.3)	**
Depression	3.5 (4.5)	3.3 (4.4)	4.8 (5.0)	**
PTSD	9.4 (11.6)	9.1 (11.6)	10.6 (11.3)	ns
Social network characteristics, m (SD)
Total number of ties	6.3 (3.8)	6.1 (3.8)	7.1 (3.8)	*
Family ties	3.1 (2.4)	3.0 (2.4)	3.9 (2.4)	**
Female ties	2.4 (2.1)	2.0 (1.7)	4.8 (2.7)	***
Close ties	4.9 (3.2)	4.7 (3.1)	6.0 (3.4)	**
Military-affiliated ties	.9 (1.2)	.9 (1.3)	1.0 (1.2)	ns
Heavy-drinking ties	.5 (.9)	.5 (1.0)	.5 (.7)	ns
Drinking-buddy ties	.8 (1.4)	.8 (1.4)	.7 (1.3)	ns
Illicit drug-using ties	.4 (.9)	.4 (1.0)	.3 (.6)	ns
Days spent with ties, past month	3.3 (3.5)	3.3 (3.6)	3.1 (2.9)	ns
Days drinking with ties, past month	1.3 (1.7)	1.4 (1.8)	1.0 (1.3)	+

SD = standard deviation; PTSD = posttraumatic stress disorder.

^aIncludes American Indian or Alaska Native, Asian, Native Hawaiian or Pacific Islander, more than one race, and other specified races.

^b“Some College” includes Trade school, Associate degrees, and other two-year technical degrees, and “College completion” includes four-year degrees and graduate degrees.

⁺p < .10,

^*p <.05,

^**p <.01,

^***p <.001

Measures

The outcomes of interest were symptom severity of the following: anger, anxiety, depression, and posttraumatic stress disorder (PTSD). Anger was assessed using the PROMIS Anger scale (Pilkonis et al., 2011), an 8-item self-rated measure of anger over the past seven days, scored on a 5-point scale (“Never”=1 to “Always”=5; Range: 8–40; Cronbach’s α=0.94). Example items include, “I was irritated more than people knew,” and “I felt angrier than I thought I should.” Anxiety symptom severity was based on the past seven days and assessed using the Severity Measure for Generalized Anxiety Disorder-Adult (Craske et al., 2013). The measure consists of 10 items asking about frequency of symptoms (“Never”=0 to “All of the time”=4), resulting in a total score range of 0 to 40. Higher scores indicate more severe generalized anxiety (Cronbach’s α=0.91). Example items include “Avoided, or did not approach or enter, situations about which I worry” and “Felt a racing heart, sweaty, trouble breathing, faint, or shaky.” Depression symptom severity was measured according to the past two weeks using the Patient Health Questionnaire 8 (PHQ-8) (Kroenke et al., 2009). The PHQ-8 is an 8-item measure that maps onto the DSM-IV diagnosis for depression and asks about frequency of depressive symptoms on a 4-point Likert scale (“Not at all”=0 to “Nearly every day”=3). Total scores range from 0 to 24, with higher scores indicating greater severity of depression (Cronbach’s α=0.91). Example items include “Little interest or pleasure in doing things” and “Feeling tired or having little energy.” PTSD was assessed based on the past 30 days using the PTSD Checklist (PCL-5), a 20-item instrument assessing how bothered a person is by their symptoms (“Not at all”=0 to “Extremely”=4). Total scores range from 0 to 80 (Cronbach’s α=0.95), and higher scores indicate greater symptom severity (Blevins et al., 2015; Bovin et al., 2015; Weathers et al., 2013).

The study examined ten characteristics of social network structure. Social network characteristics were assessed based on an existing reliable social network inventory, which has been used and adapted over the past twenty years to examine various aspects of social network composition and impacts of substance use in the network (e.g., drinking buddies and heavy drinkers) on individual outcomes (Homish & Leonard, 2008; Leonard & Homish, 2008; Leonard et al., 2000; Reifman et al., 2006). Our previous work has also demonstrated the utility of the dimensions captured in this inventory for examining substance use among USAR/NG soldiers (Anderson Goodell et al., 2020; Hoopsick et al., 2021). As part of the survey, soldiers were asked to identify as social ties individuals other than the spouse/marital partner who the soldier considered to be “important to [him or her] in one way or another during the past year” in relation to any of the following: providing emotional support, socializing regularly, helping with practical or financial problems, or providing support. Soldiers reported on up to 24 social ties each (collectively reporting a total of 2,637 ties) and answered questions relating to characteristics of interest for each tie (Leonard & Homish, 2008), capturing qualitatively different characteristics of individuals’ social network structures, including network size, composition, quality, and substance use. For analytic purposes, all characteristics were aggregated at the soldier level.

Network Size

• 1Network size – Total number of ties in the social network.

Network Composition

• 2Family member – Number of family members in the network, defined as the soldier’s child/stepchild, parent, sibling, extended family member, or in-law.
• 3Female tie – Number of females in the social network.
• 4Military affiliation – Number of ties in the network currently in the military. Identified ties could be any of the following relations to the soldier, including but not limited to: peer, supervisor, commanding officer.

Network Quality

• 5Close tie – Number of close ties in the network based on the soldier’s report of feeling “a medium amount” or “a lot” of closeness to a tie (based on the options “not at all,” “a little,” “a medium amount,” and “a lot”).
• 6Time with ties – Average number of days spent with ties in the past month.

Network Substance Use

• 7Heavy-drinking ties – Number of ties in the network who are heavy drinkers, based on the soldier’s report of a tie being either a “frequent or heavy social drinker,” “problem drinker,” or “alcoholic.”
• 8Drinking buddy – Number of ties in the network considered to be drinking buddies, i.e., a tie who the soldier “[gets] together with on a regular basis to do activities that center[…] around drinking and/or going to bars or nightclubs” (Homish & Leonard, 2008; Lau-Barraco & Collins, 2011).
• 9Illicit drug use – Number of ties in the network who engaged in any illicit drug use in the past year (according to the soldier’s report).
• 10Time drinking with ties – Average number of days spent drinking with social ties in the past month.

Other demographic and military service history characteristics were also included in the analyses as covariates at the participant level, including soldier’s age, soldier’s sex (female=1, male=0), number of years of military service, and individual deployment history (ever=1, never=0).

Statistical analyses

The current study uses data from the baseline wave of Operation: SAFETY. Descriptive statistics characterized the sample on the outcomes and predictors of interest and tests of comparisons (t-tests and Chi-square tests) examined potential differences between males and females. Separate unadjusted and adjusted negative binomial regression models for each mental health outcome (anger, anxiety, depression, and PTSD) were used to assess relationships with social network characteristics (total number of ties, family ties, female ties, close ties, military-affiliated ties, heavy-drinking ties, drinking buddies, illicit drug-using ties, days spent with ties, and days drinking with ties). Unadjusted models examined the relationship between each outcome and each social network characteristic separately (e.g., anxiety and close ties; anxiety and number of ties, depression and close ties, etc.). Adjusted models examined each mental health outcome separately, but included all ten social network characteristics and covariates (i.e., age, sex, years of military service, deployed (yes/no)) within the same model, in order to understand the relative importance of the various characteristics. The regression models used negative binomial logistic specification to account for the overdispersion in the count-based outcomes. Finally, we added interaction terms to the adjusted models to determine whether there were significant differences by sex in the relationships between social network characteristics and soldiers’ mental health outcomes.

Results

Descriptive statistics were used to characterize the sample on all variables of interest and compare males and females. Compared to female soldiers, males were older, had served longer, and were more likely to have previous deployment (Table 1). For mental health outcomes, nearly a third of the sample had at least mild depression (8.6% moderate to severe), 73% reported at least mild anxiety (13.8% moderate to extreme), and 6.7% of the sample met criteria for a probable diagnosis of PTSD (score ≥31). Female soldiers had higher scores on both the anxiety and depression measures, but did not differ from males on anger or PTSD symptoms.

Soldiers reported an average of six ties in their social networks (m=6.3; SD=3.8). Compared to male soldiers, females had a higher number of total ties in their networks, as well as more family members, female ties, and close ties. No differences were observed between male and female soldiers according to any of the alcohol or illicit drug characteristics of the social networks. There was also no difference in the average number of military ties in male and female soldier social networks.

In univariate analyses examining the individual associations between each social network characteristic and each mental health outcome separately, greater days drinking with social network ties was associated with higher levels of anger (RR: 1.02, 95% CI: 1.01, 1.04, p <.05). A greater number of social network ties who use illicit drugs was associated with higher levels of anxiety (RR: 1.15, 95% CI: 1.01, 1.31, p <.05). All other univariate relationships were non-significant (Table 2).

Table 2.

Unadjusted associations between social network characteristics and soldiers’ mental health symptom severity

	Anger	Anxiety	Depression	PTSD
	RR (95% CI)	RR (95% CI)	RR (95% CI)	RR (95% CI)
Social network characteristics
Number of ties	1.01 (1.00, 1.02)+	1.02 (.99, 1.06)	1.03 (.99, 1.06)	1.03 (.99, 1.06)
Family ties	1.01 (1.00, 1.03)+	1.04 (.99, 1.09)	1.04 (.99, 1.09)+	1.04 (.99, 1.09)
Female ties	1.01 (.99, 1.03)	1.03 (.97, 1.09)	1.05 (.98, 1.11)	1.02 (.96, 1.09)
Close ties	1.00 (.99, 1.01)	1.00 (.96, 1.04)	1.00 (.96, 1.04)	1.02 (.98, 1.06)
Military-affiliated ties	1.02 (.99, 1.05)	1.02 (.93, 1.13)	1.04 (.94, 1.15)	1.08 (.97, 1.20)
Heavy-drinking ties	1.03 (.99, 1.07)	1.04 (.91, 1.19)	1.04 (.90, 1.20)	1.06 (.92, 1.22)
Drinking buddy ties	1.02 (.99, 1.05)	1.05 (.96, 1.15)	1.03 (.93, 1.13)	1.08 (.98, 1.19)
Illicit drug-using ties	1.03 (.99, 1.07)	1.15 (1.01, 1.31)*	1.03 (.90, 1.17)	1.13 (.98, 1.30)+
Days spent w/ ties	1.00 (.99, 1.01)	1.01 (.98, 1.05)	.98 (.94, 1.01)	1.01 (.97, 1.05)
Days drinking w/ ties	1.02 (1.01, 1.04)*	1.02 (.95, 1.11)	99 (.92, 1.06)	1.06 (.97, 1.15)

RR = risk ratio; CI = confidence interval;

⁺p < .10,

^*p <.05

Multivariate analyses accounted for all social network characteristics within the same model, to examine which are most salient for contributing to mental health outcomes (Table 3). Out of all social network characteristics, number of close ties in the social network was most consistently associated with mental health symptoms, such that having more close ties in the network was associated with lower symptom severity of anger (aRR: 0.97, 95% CI: 0.96, 0.99, p<.01), anxiety (aRR: 0.93, 95% CI: 0.87, 0.99, p<.05), and depression (aRR: 0.93, 95% CI: 0.87, 0.99, p<.05), but not PTSD. Having more illicit drug-using ties in the network was associated with greater symptoms of anxiety (aRR 1.17, 95% CI: 1.01, 1.34, p <.05). Finally, more days spent drinking with social network members was related to higher levels of anger (aRR: 1.02, 95% CI: 1.01, 1.05, p <.05). Taken together these results suggest that close social networks are associated with fewer mental health symptoms, while greater substance use in the social network is associated with greater mental health symptoms. There were no significant interactions between soldier sex and any of the social network characteristics.

Table 3.

Multivariate associations between social network characteristics and soldiers’ mental health symptom severity

	Anger	Anxiety	Depression	PTSD
	aRR (95% CI)	aRR (95% CI)	aRR (95% CI)	aRR (95% CI)
Social network characteristics
Number of ties	1.01 (.99, 1.03)	1.07 (.99, 1.15)+	1.06 (.97, 1.14)	1.03 (.95, 1.11)
Family ties	1.02 (1.00, 1.04)	1.03 (.96, 1.11)	1.03 (.95, 1.11)	1.03 (.95, 1.12)
Female ties	1.00 (.97, 1.03)	.95 (.86, 1.05)	.98 (.88, 1.09)	.97 (.86, 1.08)
Close ties	.97 (.96, .99)**	.93 (.87, .99)*	.93 (.87, .99)*	.96 (.90, 1.03)
Military-affiliated ties	1.02 (.98, 1.05)	.98 (.87, 1.11)	1.01 (.89, 1.14)	1.02 (.90, 1.16)
Heavy-drinking ties	1.01 (.97, 1.05)	.96 (.82, 1.12)	1.01 (.86, 1.19)	.99 (.85, 1.15)
Drinking buddy ties	1.01 (.98, 1.04)	1.05 (.94, 1.17)	1.01 (.90, 1.13)	1.05 (.94, 1.19)
Illicit drug-using ties	1.04 (1.00, 1.08)+	1.17 (1.01, 1.34)*	1.06 (.92, 1.22)	1.16 (1.00, 1.35)+
Days spent w/ ties	1.00 (.99, 1.01)	1.01 (.97, 1.05)	.98 (.94, 1.02)	1.00 (.96, 1.04)
Days drinking w/ ties	1.02 (1.01, 1.05)*	1.02 (.94, 1.11)	1.02 (.94, 1.10)	1.04 (.95, 1.14)
Soldier characteristics
Age	.99 (.98, 1.00)+	.97 (.94, .99)*	.96 (.93, .99)*	.96 (.93, .99)**
Sex (male = ref)	1.12 (1.00, 1.26)+	1.79 (1.20, 2.68)**	1.82 (1.18, 2.80)**	1.46 (.95, 2.25)+
Years serving	1.01 (1.01, 1.02)**	1.06 (1.02, 1.10)**	1.05 (1.01, 1.09)*	1.08 (1.04, 1.11)***
Ever deployed (no = ref)	.98 (.90, 1.07)	.93 (.70, 1.25)	1.21 (.89, 1.64)	1.12 (.82, 1.54)

aRR = adjusted risk ratio; CI = confidence interval;

⁺p <.10,

^*p <.05,

^**p <.01,

^***p <.001

Discussion

Using data on R/NG soldiers’ mental health and social networks, the primary aim of the study was to understand the relative importance of a variety social network characteristics on symptom severity of anger, anxiety, depression, and PTSD. Among all social network characteristics examined, the number of close ties, the number of illicit drug-using ties, and days drinking with ties were important for soldiers’ mental health. Specifically, having more close ties in the network was associated with lower symptoms scores for anger, anxiety, and depression, but not PTSD. While evidence from other studies has demonstrated network size and diversity to be consistently important for mental health (Hatch et al., 2013; Platt et al., 2014; Rueger et al., 2016), our results indicate that among this group of R/NG soldiers, size of the network may be less important than the number of perceived close social relationships. This is consistent with another study of National Guard soldiers, which demonstrated that both general social support and support within the military unit were associated with lower odds of having a mental health condition, while size and diversity of relationship types were not (Sripada et al., 2015). This may be a result of different social network configurations among R/NG soldiers, which may render close ties more important than a broad/diverse network that may not understand their unique dual civilian and military belonging. Interestingly, time spent with ties was not significantly related to mental health outcomes in our data, indicating that knowledge of having close ties in an individuals’ network may be more important than the amount of time actually spent together.

Further, based on previous literature, we expected having military ties in the network may be important for understanding mental health symptoms among R/NG soldiers. Contrary to this expectation, the number of military-affiliated ties was not associated with symptom severity for any of the mental health outcomes in our sample. Previous work has demonstrated mixed effects of military networks. Among UK military members, having a network of mostly other military members was related to greater alcohol misuse, and increased odds of common mental disorders and PTSD among former soldiers (Hatch et al., 2013). This may be explained by former soldiers feeling disconnected from their military networks after leaving service, and having formed limited connections with non-military peers as veterans. Contrary to this, we have previously demonstrated that military ties are protective against alcohol misuse among R/NG soldiers, although the relationship was observed specifically in deployed soldiers (Anderson Goodell et al., 2020). This may be a result of the unique features of R/NG service and soldiers’ overlapping civilian and military networks, both of which may be equally salient for these individuals, in a way that they may not be for active duty military members, for whom military networks have primary importance (Griffith, 2009). This has important implications for interventions; contrary to the common assumption that veterans feel most supported by other veterans, our findings suggest that social network ties for R/NG soldiers may not necessarily have to be military-affiliated to offer some benefit.

To our knowledge, this is the first examination of social network characteristics among military service members that has examined the relationship between these characteristics and anger. Specifically, our data indicate that having a greater number of close ties in the social network is associated with less symptoms of anger. While anger has received less attention overall as a mental health concern of interest, new evidence indicates it may be critical for understanding other mental health problems (Adler et al., 2020; Dillon et al., 2021; Dillon et al., 2020; Worthen et al., 2014). In particular anger is related to PTSD (Worthen et al., 2014), suicidal ideation (Dillon et al., 2020), non-suicidal self-injury (Dillon et al., 2021), and other outcomes (Adler et al., 2020) among military service members. Evidence from the Millennium Cohort Study, an ongoing study of all branches of the U.S. military, also indicates that anger may increase as individuals leave military service (Adler et al., 2020). In this context our findings have important implications for addressing anger among military service members. Facilitating close social relationships may be an important strategy to help minimize the effects of anger and related sequelae. Future studies are needed to examine the temporal and causal relationships between anger, close social relationships, and other mental health outcomes.

Finally, our results also show the importance of considering social network substance use in relation to soldier mental health outcomes. Specifically, having more social ties who used illicit drugs in the past year was associated with increased symptoms of anxiety, and spending more days drinking with ties was associated with increased symptoms of anger. To our knowledge, there has been limited examination of the impacts of substance use in the social network on outcomes other than substance use, in both the general population and specifically within the military. Our results demonstrate this may be an important future consideration. Future research should focus on a more granular understanding of the relationship between substance use characteristics of the social network and mental health symptoms, specifically for anxiety and anger. The current work does not establish causal pathways between the measures of interest, so it is possible that having social ties who use substances may influence a soldier with mental health symptoms to consider using, or that individuals who have mental health symptoms may look to substance use for self-medication and then gain ties in their network who also use. Further, different categories of substance use among the network may be impacted in different ways by military participation. For example, due to the prominent culture of alcohol use in the military (Ames & Cunradi, 2004; Ames et al., 2007), time spent drinking with ties may be more accepted among military members of the network, while use of other substance, such as illicit drugs, may be more prominent in civilian networks. How these differences may relate to differing substance use motivations (e.g., social vs. self-medication) and mental health outcomes is unclear. Future work that examines causal pathways, and interactions between network substance use, military versus civilian social groups, and mental health is an important next step. In addition, future work might also examine the types of illicit drugs that social ties typically use, given that use of certain substances may negatively affect mental health outcomes, as has been shown in specifically military work connecting marijuana use to increased PTSD symptoms (Allan et al., 2019).

Our secondary aim was to examine whether any of these relationships differed based upon the sex of the soldier. All interaction terms between soldier sex and the social network characteristics were not significant, indicating that social network factors impact mental health in similar ways for both male and female soldiers in spite of the fact that female soldiers had a higher number of total ties, more family members, more female ties, and more close ties compared to male soldiers. Ample other evidence indicates that social network characteristics and the ways in which they relate to mental health may differ for men and women (Homish & Leonard, 2008; Kawachi & Berkman, 2001). Contrary to those findings, which were primarily among civilian samples, our results indicate that while social network characteristics may differ between male and female soldiers, the relationships of these characteristics to mental health outcomes do not. Replication of these findings in a larger sample of female soldiers is an important next step for determining the extent to which social support interventions for military service members need to be tailored based on gender.

The results of this study should be considered in light of limitations. First, due to the aims of the overall study, the sample was limited to soldiers who were married/partnered at baseline. As such, results may not generalize to soldiers who are single, as their social networks may be different. However, it is important to note that just under half of USAR/NG soldiers are or were recently married (Department of Defense (DoD), 2021). Second, soldiers’ recall of all ties in their social networks may be subject to bias. While it is not possible to confirm whether soldiers named all possible important ties from memory, research shows that individuals are less likely to forget strong ties, characterized by closeness, reciprocity, and recency and frequency of contact (Brewer, 2000). To help encourage recall of important ties, the survey included prompts for soldiers to list individuals who were important to them for different reasons, including emotional, practical, and social reasons. Additionally, the data did not include detailed information to contextualize the soldiers’ closeness to social ties. Other measures examining constructs like emotional and tangible support would have been helpful to understand specific elements of closeness that may be related to lower mental health symptoms (Cohen et al., 1985; Cutrona & Russell, 1990). Third, levels of mental health symptomatology were based on self-report, rather than clinical assessment and were on average below diagnostic thresholds and the clinical relevance of these findings needs to be examined in greater detail. It is important to note however, that this was a descriptive epidemiological study of a community, rather than clinical sample, and there were still significant proportions of participants reporting clinically relevant symptoms (e.g., 13.8% moderate to extreme anxiety). Further, even subthreshold mental health symptoms have been shown to be important for future mental and physical health outcomes (Dillon et al., 2021; Pietrzak et al., 2021). The clinical significance of the associations identified between these specific social network characteristics and mental health symptomatology also needs to be further established in order to understand the overall importance of the social network compared to other individual-level factors and inform practical interventions. However, it is important to note that social relationships are already a well-established factor in mental health (Thoits 2011), and our results provide additional insight into potentially important aspects of these relationships to consider. Finally, the data for the current work are cross-sectional, so the results are unable to distinguish temporality from the observed associations. It is possible that social network composition could lead to symptom severity (i.e., social causation), or symptom severity could dictate social network composition (i.e., social selection) (Johnson, 1991; Kaniasty & Norris, 2008; Steglich et al., 2010). Our goal was to assess potential associations, and longitudinal research should be pursued to clarify temporality. Strengths of this work hinge on the rich data set resulting from a robust social network inventory that was used to estimate the relationships between social network characteristics and symptoms of anger, depression, anxiety, and PTSD.

In conclusion, this study contributes knowledge about social network characteristics that may influence military mental health outcomes and indicates that the quality of U.S. R/NG soldiers’ mental health may be impacted by prosocial interpersonal relationships as well as social network substance use behaviors. Specifically, the results suggest that interpersonal relationships that entail substance use are associated with greater symptoms of anxiety and anger while those that reflect closeness between the tie and soldier are associated with fewer symptoms of anger, anxiety, and depression, regardless of whether or not the tie has military experience. While the clinical significance of these associations needs to be further established, our findings add to an understanding of the complex social-ecological factors that contribute to individual mental health. This underscores the need for holistic, whole person approaches to mental health, which consider not only the individual, but their broader social context. As other researchers have noted (e.g., Reifman 2006), while we cannot necessarily change individual’s social networks, our findings have implications for the development of interventions to ameliorate the potential negative mental health impacts of military service. Social network composition may be an educational intervention point whereby individuals’ can learn to be aware of the ways in which their networks impact their well-being (Reifman 2006), in both positive and negative ways. These findings call attention to the need to further understand relationships between social networks and soldiers’ symptom severity, and the relative clinical significance of social network factors, so as to optimize mental health outcomes.