Perceived friendships protect against the development of anger following childhood adversities in UK military veteran men residing in Northern Ireland

ABSTRACT

Background: Experience of childhood adversity is associated with greater anger as an adult, particularly in men. Soldiers and veterans report higher incidence of adverse childhood experiences, many of whom also experience elevated rates of PTSD and anger. However, little is known about factors which may protect against the development of anger after experiencing childhood adversity.

Objective: This study aims to assess the potential protective aspects of perceived social support in military veterans.

Methods: Data from the Northern Ireland Veterans’ Health and Wellbeing Study (N = 590, Mage = 56) was utilised in regression models to examine perceived social support (family, friend, partner; MSPSS) as a moderator of the association between adverse childhood experiences (ACEQ-10) and anger (DAR-7). This sample comprised men who were UK Armed Forces veterans residing in Northern Ireland.

Results: Significant interaction effects, visualised using interaction plots, were found between perceived friend support and both child abuse and household challenge. When men perceived high friend support, there was no association between child abuse or household challenge and anger. For veteran men who perceived the maximum amount of partner support, there was no association between child abuse and anger. Family support did not change the positive association between child abuse, child neglect or household challenge and future anger.

Conclusions: This study indicates that it is especially important to foster supportive and empathetic friendships for men that have experienced adversity as a child, perhaps through programmes such as Men’s Sheds, as these friendships may alleviate the negative influences of child abuse and household challenge on anger.

HIGHLIGHTS

• Perceived friend and partner support were protective factors against the development of anger for veterans that experienced child abuse or household challenge.

• Child neglect had the strongest association with adult anger, unchanged by any type of perceived support.

• Programmes which foster friendships may be particularly beneficial for veterans transitioning out of the military.

Antecedentes: Experiencias infantiles adversas, están asociadas con un mayor sentimiento de ira durante la adultez, especialmente en hombres. Soldados y veteranos de guerra, han reportado mayor incidencia de experiencias infantiles adversas, muchos de las cuales también han presentado altas tasas de trastorno de estrés post traumático (TEPT) y problemas de ira. Sin embargo, poco se conoce sobre los factores protectores que podrían prevenir el desarrollo de problemas de ira, después de haber experimentado eventos adversos durante la niñez.

Objetivo: Este estudio tiene como objetivo evaluar los potenciales aspectos protectores recibidos como ayuda social en militares veteranos.

Método: Datos recolectados desde el Estudio de salud y bienestar de veteranos de Irlanda del Norte (N = 590, edad media = 56) fueron utilizados en un modelo regresivo para examinar la percepción del apoyo social (familia, amigos, pareja; MSPSS) como un moderador entre las experiencias infantiles adversas (ACEQ-10) y la ira (DAR-7). La muestra comprendía a hombres residentes en Irlanda del Norte, quienes fueron veteranos de las Fuerzas Armadas del Reino Unido.

Resultados: Se encontraron efectos de interacción significativos, visualizados en modelos de interacción, entre el apoyo percibido de un amigo y tanto el abuso infantil como el desafío doméstico. Cuando los hombres percibieron un alto apoyo de sus amigos, no existió asociación entre el abuso infantil y el desafío doméstico y la ira. En el caso de los veteranos que percibieron el máximo apoyo de sus parejas, no existió asociación entre el abuso infantil y la ira. El apoyo familiar no cambió la asociación positiva entre el abuso infantil, la negligencia o el desafío doméstico y los problemas de ira futuros.

Conclusiones: Este estudio indica que es especialmente importante fomentar las amistades empáticas y de apoyo para los hombres que han experimentado eventos adversos durante la niñez, quizás a través de programas como Men’s Sheds (Comunidad), ya que estas amistades pueden atenuar las influencias negativas del abuso infantil y el desafío doméstico sobre la ira.

KEYWORDS:

• Veterans
• anger
• childhood adversities
• social support
• friendship

PALABRAS CLAVES:

• Veteranos
• ira
• adversidades infantiles
• apoyo social
• amistad

1. Introduction

Anger can become problematic depending on the frequency, duration, and intensity in which it occurs, and the level of impact it has on general and social functioning (Novaco et al., 2012). Adler et al. (2020) found an increase in problematic anger during the months before and after military separation in a cohort study of 3,448 service members transitioning out of the military. However, rates of problematic anger almost doubled 24 months after military separation (31%), when compared to rates of problematic anger 24 months pre-separation (16%). Increased rates of problematic anger were associated with behavioural, functional, relationship, and economic problems five years later, including increased rates of depression and PTSD. Further, problematic anger has consistently been linked to suicide attempts in military populations, particularly in veterans rather than active military personnel (Naifeh et al., 2021; Varker et al., 2022).

Problematic anger exhibits high prevalence (10–31%) in trauma-affected populations (Forbes et al., 2022). This is particularly relevant for military populations, as these individuals are likely to experience trauma during deployment. Further, veterans self-report a higher incidence of adverse childhood experiences (ACEs; Katon et al., 2015). The ACEs study (Felitti et al., 1998) highlighted the long-term impact of experiencing multiple child adversities on physical health problems, depression, and suicide attempts. Thus, veterans are a high-risk population regarding potential maladaptive outcomes, such as problematic anger, associated with experiencing childhood adversity as well as military-related violence and trauma.

Research on ACEs with military personnel found that men in the military reported an average of 1.6 ACEs, whereas civilian men reported an average of 1.3 ACEs (Katon et al., 2015). Using data from the Northern Ireland (NI) Study of Health and Stress, McLafferty et al. (2015) identified the prevalence of childhood adversities in adults residing in Northern Ireland as 6.1% parent mental illness, 5.4% family violence, 3.7% physical abuse, 2.3% sexual abuse, 2.2% parent substance abuse, and 1.9% neglect. Recently, research has been conducted to look at the reported rates of ACEs in UK military veterans residing in NI through the Northern Ireland Veterans’ Health and Wellbeing Study. Reported rates of childhood adversities in the 656 veterans included in the study were 20.5% parent mental illness, 14.8% family violence, 32.8% physical abuse, 12.9% sexual abuse, 22.7% parent substance abuse, and 12.1% neglect (McLafferty et al., 2021). Comparing these two studies suggests that adults who served in the military are more likely to have experienced adversity as a child, particularly physical abuse; however, comparisons need to be made cautiously given one study was representative of the population (community adults) and the other was not (military veterans). It is further important to note that certain ACEs, such as child abuse, lead to more negative outcomes than others as not all ACEs are equal (Merrick et al., 2017).

Depending on the number and type of ACEs, adults may develop greater resilience (Zautra et al., 2010) or a range of negative outcomes, including physical and mental ill health (Chapman et al., 2004; Felitti et al., 1998). In a community sample of adults in Ireland, experience of child emotional and physical abuse and neglect, rather than household dysfunction, was most strongly related to later-life depressive symptoms (Von Cheong et al., 2017). Military veterans in NI who experienced multiple ACEs were 2.5–3 times more likely to have mental health problems, and those who experienced child abuse were more likely to have anxiety or depression, when compared to military veterans who reported low/no rates of adverse experiences (McLafferty et al., 2021). Further, NI military veterans who were most likely to endorse all ACEs, when compared to those who reported different patterns of ACEs (low adversity, chaotic home environment, and physical and psychological abuse latent classes as reported by McLafferty et al., 2021), had a significantly elevated risk for post-military incarceration (McGlinchey & Armour, 2023). Anger is also associated with higher ACEs, particularly in men (Messina & Schepps, 2021), and is self-reported as the greatest PTSD-related concern for veterans in treatment (Rosen et al., 2013). Apart from being a core concern in PTSD treatment, anger has been highlighted as an issue even for veterans and soldiers without a PTSD diagnosis due to the ‘institutionalized role anger plays in military training, identity, and culture’ (Miller, 2019, p. 7).

Prolonged feelings of anger may increase the risk for heart problems (Chida & Steptoe, 2009), and lead to poor psychosocial functioning and anti-social behaviours (Bettencourt et al., 2006). However, during active military duty, anger can serve a functional purpose, as it is associated with a readiness for engagement (Forbes et al., 2022). This may be particularly useful in situations where it is necessary to quickly engage with and neutralise hostile threats. Outside of active military duty, anger becomes problematic when it is associated with continuous hostile appraisal of situations. Research conducted with Australian and U.S. military personnel indicates that problematic anger increases from 17% during active duty to 30% in veterans or transitioned personnel (Adler et al., 2020; Forbes et al., 2018). The increase in rates of problematic anger in veterans, and associated negative mental and physical health outcomes, highlight the importance of studying factors which may mitigate the development of problematic anger.

It has been noted in extant research that perceived social support moderates the association between experience of child abuse and later-life depression, such that there was no association between child abuse and depression for adults who reported high perceived social support (Von Cheong et al., 2017). Perceived social support concerns an individual’s perception of the availability and quality of support received from their social network and is linked to positive social well-being and mental health (Haber et al., 2007). In particular, the perception of support received, rather than the objective received support ratings, is most strongly associated with psychological distress following stressful events (Sarason et al., 2001). Higher perceived social support influences cognitive appraisals of the behaviour of others as more helpful and supportive (Lakey & Cassady, 1990), and acts as a stress-buffer against pathogenic effects of negative life events (Cohen et al., 2001).

In a meta-analysis of PTSD risk factors, lack of social support was the second most important PTSD risk factor, behind trauma severity, for military personnel (Brewin et al., 2000). Wilks et al. (2019) found that high perceived social support protected against suicidal ideation for veterans who reported both high and low levels of anger. Perceived social support may therefore protect against the maladaptive influence of childhood adversities on future anger, particularly in military veterans who are likely to experience additional combat-related trauma. The source of perceived support, such as from friends, family, or a partner or spouse, may also have varying influences on the association between child adversity and anger.

In the context of childhood adversity, certain family members may have either been seen as a source of abuse or support. If family is the source of abuse, friend and partner support may be even more necessary to achieve healthier long-term outcomes. Family and friends may also be an ingrained, long-term source of support and thus have historical effects on perceived social support. Perceived friend support may be particularly salient for military personnel, as soldiers often form close-knit friendships within military units. Evans et al. (2013) examined different types of perceived social support, including family and friend support, with adults who experienced childhood adversity. Results indicated both high perceived friend and family support predicted lower total trauma symptoms for men who experienced various types of child abuse or neglect. In a community sample of young adults, only perceived friend support, rather than family or partner support, buffered the association between perceived stress and loneliness (Lee & Goldstein, 2016). Inclusion of both specific social support systems (i.e. family, friends, and partner) and types of childhood adversity (i.e. abuse, neglect, and household challenge) will enable us to identify which types of social support have specific or general benefits across a range of adverse childhood experiences for military veterans.

The current study will examine perceived friend, partner, and family social support as moderators of the association between ACEs (IVs) and self-reported anger (DV). Rather than a total ACE score, three separate scores of child abuse, neglect, and household challenge (IVs) will be included as predictors of anger (DV). It is hypothesised that child abuse and neglect will have strong associations with future anger, and that perceived social support will moderate this relationship. The current study is also exploratory in nature as it will be the first to examine the influence of different sources of perceived support (i.e. friend, family, and partner) on military veterans.

2. Methods

2.1. Design

The Northern Ireland Veterans Health and Wellbeing Study (NIVHWS) included a cross-sectional psychological wellbeing survey implemented between December 2017 and June 2019. The survey was advertised through social media channels, local service providers, newspapers, and via attendance of the research team at local events for UK Armed Forces veterans residing in NI. The survey was available online via the use of Qualtrics but also as a pen-and-paper version; however, most participants responded online. The survey comprised a wide range of modules covering sociodemographics, health, wellbeing, and behavioural concerns. Ethical approval was provided by Queen’s University Belfast’s Engineering and Physical Sciences Faculty Research Ethics Committee (EPS 19_156). In total 1,328 participants provided written consent and completed the survey. Participants were allowed to skip questions if they wished, and as such, there were large amounts of missing data.

2.2. Missing data

Participants identifying as a gender other than male were removed from the sample (N = 159 removed). Of the remaining sample, 584 men had no missing responses, 112 were missing between 1 and 20 items, and the remaining 473 were missing all 29 items of the key variables of interest: social support, anger, and ACEs. Missing values were most common on perceived social support, with 581 participants missing some or all items. The measure of ACEs had the least missing values, with 478 participants missing some or all items. Participants were removed if they had 100% missing item-level data for any of the variables of interest (N = 579 removed). T-tests were used to examine significant differences between participants who were included and excluded. There was not a significant difference in age or years of military service in regular and reserve forces. The average number of ACEs was 2 for both included and excluded participants. Included participants (M = 4.2, SD = 2.41), compared to excluded participants (M = 3.69, SD = 2.27), reported a significantly higher level of education, t(1001) = 3.50, p = .0005. For included participants, missing data was missing completely at random (MCAR) according to Little’s (1988) MCAR test, which was conducted using mcar_test in R Studio, χ²(87) = 84.01, p = .57. Data was also MCAR for excluded participants, χ²(91) = 88.29, p = .56. As there was only a small amount of missing data in the remaining included sample (6 participants with varying levels of missing data), models were run according to pairwise inclusion, in which available data is used for specific models. This method has been found to be equivalent to multiple imputation, even with small sample size, low association among items, and small number of items per variable (Parent, 2013). Therefore, the final sample used was 587–590, depending on the variables included in the specific regression model.

2.3. Sample

The effective sample for this study comprised 590 men who were UK Armed Forces veterans residing in Northern Ireland. Mean age of the sample was 56 years (Range = 25–99 years) and 99.3% were White. Regarding current relationship status, 75.4% indicated that they were married or living with a partner, 16.3% were separated or divorced, and 4.6% were single or never married. Of the 588 participants who reported their education status (2 missing), 42% completed at least a Certificate of Higher Education or NVQ Level 4, and 19% left school with no qualifications.

2.4. Measures

2.4.1. Adverse childhood experiences (ACEs)

Number and type of ACEs before the age of 18 were self-reported using the 10-item Adverse Childhood Experiences Questionnaire (ACE-Q; Felitti et al., 1998). These 10 items can be grouped into three adverse categories: child abuse (emotional, physical, and sexual abuse; Range = 0–3), child neglect (physical and emotional neglect; Range = 0–2), and household challenges (family mental illness, parent treated violently, divorce, incarcerated parent, parent substance abuse; Range = 0–5; Centers for Disease Control and Prevention, 2021). Participants responded ‘yes’ or ‘no’ to each of the 10 questions.

2.4.2. Perceived social support

Participants rated perceived social support from their family, friends, and partner on a 1–7 Likert scale (1 = Very Strongly Disagree) with the Multidimensional Scale of Perceived Social Support (MSPSS; Zimet et al., 1988). The MSPSS is a 12-item scale, with four items each measuring family (Cronbach’s α = .92), friend (α = .92), and partner (α = .95) support. An average total perceived social support score can be obtained, as well as a score for each type of support (Range = 1–7).

2.4.3. Anger

Participants self-reported on anger using the Dimensions of Anger Reactions – 7 Scale (DAR-7; Novaco, 1975). The DAR-7 measures anger response and anger impairment and generates a total anger score. A 0–4 Likert scale (0 = Not at all) was used for response options (Hawthorne et al., 2006). Only the total anger score (Range = 0–28; Cronbach’s α = .95) was used in the current study. Validation studies in Spain and Australia suggest a cut-off score of 10 or 8, respectively, for problematic anger (Kannis-Dymand et al., 2019).

2.5. Statistical analyses

All analyses were conducted in R Statistical Software package (R Core Team, 2020). No variables had skewness or kurtosis values beyond acceptable cut-offs. Predictors (child abuse, neglect, and household challenge) and moderators (family, friend, and partner support) were centred (i.e. subtracting sample means to produce a new variable with a mean of zero) to control for multicollinearity among the predictor, moderator, and interaction term (predictor x moderator; Frazier et al., 2004) prior to running the regression model to predict adult anger (DV). Moderation effects were identified using regression models and analysis of variance tests. Significant interaction effects were examined with interaction plots with moderators split at ±1 standard deviation. For the models which examined perceived partner support as a moderator, data was filtered to only include men who were married or living with a partner (household challenge model: N = 439; abuse and neglect models: N = 442).

3. Results

3.1. Descriptive statistics

In this sample of male veterans drawn from the NIVHWS, 43.2% experienced at least one type of child abuse, 48.6% experienced at least one type of household challenge, and 30.8% experienced at least one type of neglect. 27.6% of participants reported anger scores greater than the cut-off of 10 on the DAR-7. Anger was significantly correlated with age (r = −0.23) and level of education (r = −0.18), so these variables were controlled for in all regression models. Correlations between all variables are presented in Table 1.

Table 1. Correlations of anger, perceived support, ACEs, and demographic characteristics.

	Age	Educ	Anger	PartSS	FamSS	FriSS	CAb	CNe
Age	–
Educ	−0.04	–
Anger	−0.23	−0.18	–
PartSS	0.10*	0.03	−0.25	–
FamSS	0.21	0.07	−0.28	0.71	–
FriSS	0.14	0.15	−0.31	0.55	0.55	–
CAb	−0.12**	−0.01	0.23	−0.07	−0.20	−0.13**	–
CNe	−0.12**	−0.10*	0.29	−0.18	−0.25	−0.25	0.60	–
HCh	−0.19	−0.06	0.16	−0.08*	−0.14**	−0.13**	0.44	0.46

Notes: Correlations greater than 0.14 are significant at p < .001; *p < .05, **p < .01. Educ = Education, PartSS = Partner Social Support, FamSS = Family Social Support, FriSS = Friend Social Support, CAb = Child Abuse, CNe = Child Neglect, HCh = Household Challenge.

3.2. Child adversity, perceived friend support, and adult anger

3.2.1. Child abuse

The regression model including age, education, child abuse, and perceived friend support was significant and explained 19% of the variance in adult anger (Table 2). Age, education, and perceived friend support were negative predictors, whereas child abuse was positive. There was a significant interaction effect between child abuse and perceived friend support (Figure 1). When veterans reported high perceived friend support (+1 SD), there was no association between child abuse and anger. However, when veterans reported low perceived friend support (−1 SD), there was a significant positive association between child abuse and anger.

Figure 1. Perceived friend social support (SS) moderates the association between child abuse and adult anger. F(5, 579) = 26.91, p < .001, R² = 0.19; F$\mathrm{\Delta }$(1, 578) = 4.94, p < .027. Low Friend SS: b = 1.97, SE = 0.40, p < .001; High Friend SS: b = 0.62, SE = 0.46, ns.

Table 2. Age, education, child adversity, and perceived friend support as predictors of anger.

	Step 1			Step 2
Predictors	B	SE B	Β	B	SE B	β
Age	−0.13	0.03	−0.18***	−0.13	0.03	−0.18***
Educ	−0.49	0.12	−0.15***	−0.50	0.12	−0.15***
CAb	1.38	0.30	0.17***	1.29	0.31	0.16***
FriSS	−1.24	0.20	−0.24***	−1.24	0.20	−0.24***
CAbxFriSS				−0.45	0.20	−0.08*
Child	R^2 = .18			R^2 = .19
Abuse	F(4, 579) = 32.18***			F(5, 578) = 26.91***
				F$\mathrm{\Delta }$(1, 578) = 4.94*
Age	−0.13	0.03	−0.18***	−0.13	0.03	−0.18***
Educ	−0.45	0.12	−0.14***	−0.45	0.12	−0.14***
CNe	2.43	0.47	0.20***	3.34	1.40	0.28***
FriSS	−1.11	0.21	−0.21***	−1.02	0.24	−0.19***
CNexFriSS				−0.23	0.33	−0.08
Child	R^2 = .19			R^2 = .19
Neglect	F(4, 579) = 34.04***			F(5, 578) = 27.30***
				F$\mathrm{\Delta }$(1, 578) = 0.48, ns
Age	−0.13	0.03	−0.19***	−0.13	0.03	−0.19***
Educ	−0.46	0.13	−0.14***	−0.46	0.13	−0.14***
HCh	0.56	0.27	0.08*	0.47	0.27	0.07
FriSS	−1.31	0.21	−0.25***	−1.32	0.21	−0.25***
HChxFriSS				−0.39	0.18	−0.08*
Household	R^2 = .16			R^2 = .17
Challenge	F(4, 576) = 27.31***			F(5, 575) = 22.96***
				F$\mathrm{\Delta }$(1, 575) = 4.85*

Notes: Presented values are from Steps 1 and 2 of the regression models with perceived friend support as a moderator of the associations between childhood adversities and adult anger. Educ = Education, CAb = Child Abuse, FriSS = Friend Social Support, CNe = Child Neglect, HCh = Household Challenge. *p < .05, ***p < .001.

3.2.2. Child neglect

The regression model including age, education, child neglect, and perceived friend support was significant and explained 19% of the variance in anger (Table 2). Main effects matched those in the child abuse model. The interaction effect between child neglect and perceived friend support was not significant. There was a positive association between child neglect and adult anger at high and low levels of perceived friend support.

3.2.3. Household challenge

The regression model including age, education, household challenge, and perceived friend support was significant and explained 17% of the variance in anger (Table 2). Main effects were the same as in the child abuse model. There was a significant interaction between household challenge and friend support (Figure 2). There was no association between household challenge and anger for veterans that perceived high friend support (+1 SD). There was a significant positive association between household challenge and anger for veterans that perceived low friend support (−1 SD).

Figure 2. Perceived friend social support (SS) moderates the association between household challenge and adult anger. F(5, 575) = 22.96, p < .001, R² = 0.17; F$\mathrm{\Delta }$(1, 575) = 4.85, p < .029. Low Friend SS: b = 1.05, SE = 0.35, p < .003; High Friend SS: b = −0.12, SE = 0.41, ns.

3.3. Child adversity, perceived partner support, and adult anger

3.3.1. Child abuse

The regression model including age, education, child abuse, and perceived partner support was significant and predicted 19% of the variance in adult anger for veterans who were married or living with a partner (Table 3). Child abuse was a positive predictor, whereas age, education, and perceived partner support were negative predictors. There was a significant interaction effect between child abuse and perceived partner support (Figure 3). However, this was only true when adults reported the maximum perceived partner support (Max = 7). When veterans reported very high partner support, child abuse was not associated with anger. When veterans reported low to average partner support, child abuse was positively associated with anger.

Figure 3. Perceived partner social support (SS) moderates the association between child abuse and adult anger. F(5, 436) = 19.80, p < .001, R² = 0.19; F$\mathrm{\Delta }$(1, 436) = 5.94, p < .016. Low Partner SS: b = 2.80, SE = 0.61, p < .001; Max Partner SS: b = 0.79, SE = 0.48, ns.

Table 3. Age, education, child adversity, and perceived partner support as predictors of anger.

	Step 1			Step 2
Predictors	B	SE B	β	B	SE B	β
Age	−0.13	0.03	−0.18***	−0.13	0.03	−0.18***
Educ	−0.64	0.14	−0.20***	−0.63	0.14	−0.19***
CAb	1.58	0.35	0.20***	1.49	0.35	0.19***
PartSS	−0.99	0.23	−0.19***	−0.90	0.23	−0.18***
CAbxPartSS				−0.50	0.21	−0.11*
Child	R^2 = .17			R^2 = .19
Abuse	F(4, 437) = 23.00***			F(5, 436) = 19.80***
				F$\mathrm{\Delta }$(1, 436) = 5.94*
Age	−0.12	0.03	−0.17***	−0.12	0.03	−0.17***
Educ	−0.56	0.14	−0.17***	−0.57	0.14	−0.17***
CNe	2.99	0.54	0.25***	2.85	0.54	0.24***
PartSS	−0.82	0.23	−0.16***	−0.76	0.23	−0.15**
CNexPartSS				−0.55	0.33	−0.07
Child	R^2 = .19			R^2 = .20
Neglect	F(4, 437) = 26.02***			F(5, 436) = 21.44***
				F$\mathrm{\Delta }$(1, 436) = 2.72, ns
Age	−0.13	0.03	−0.18***	−0.13	0.03	−0.18***
Educ	−0.58	0.15	−0.18***	−0.59	0.15	−0.18***
HCh	0.75	0.32	0.11*	0.70	0.32	0.10*
PartSS	−1.02	0.23	−0.20***	−0.94	0.24	−0.18***
HChxPartSS				−0.26	0.19	−0.06
Household	R^2 = .15			R^2 = .15
Challenge	F(4, 434) = 18.54***			F(5, 433) = 15.26***
				F$\mathrm{\Delta }$(1, 433) = 1.95, ns

Notes: Presented values are from Steps 1 and 2 of the regression models with perceived partner support as a moderator of the associations between childhood adversities and adult anger. Educ = Education, CAb = Child Abuse, PartSS = Partner Social Support, CNe = Child Neglect, HCh = Household Challenge. *p < .05, **p < .01, ***p < .001.

3.3.2. Child neglect

The model including age, education, child neglect, and perceived partner support was significant and explained 20% of the variance in anger (Table 3). Main effects matched those in the child abuse model. There was not a significant interaction between child neglect and perceived partner support, such that the positive association between child neglect and adult anger was unmoderated by perceived partner support.

3.3.3. Household challenge

The model including age, education, household challenge and perceived partner support was significant and explained 15% of the variance in anger (Table 3). Main effects were the same as in the child abuse model. The interaction effect between household challenge and perceived partner support was not significant.

3.4. Child adversity, perceived family support, and adult anger

3.4.1. Child abuse

The model including age, education, child abuse, and perceived family support was significant and explained 17% of the variance in adult anger (Table 4). Age, education, and family support were negative predictors of anger, whereas child abuse was positive. There was not a significant interaction effect between child abuse and perceived family support.

Table 4. Age, education, child adversity, and perceived family support as predictors of anger.

	Step 1			Step 2
Predictors	B	SE B	β	B	SE B	β
Age	−0.13	0.03	−0.18***	−0.13	0.03	−0.18***
Educ	−0.58	0.12	−0.18***	−0.58	0.12	−0.18***
CAb	1.36	0.31	0.17***	1.37	0.31	0.17***
FamSS	−0.95	0.19	−0.19***	−0.96	0.20	−0.20***
CAbxFamSS				0.03	0.18	0.01
Child	R^2 = .17			R^2 = .17
Abuse	F(4, 578) = 29.20***			F(5, 577) = 23.33***
				F$\mathrm{\Delta }$(1, 577) = 0.02, ns
Age	−0.13	0.03	−0.18***	−0.12	0.03	−0.17***
Educ	−0.52	0.12	−0.16***	−0.50	0.12	−0.15***
CNe	2.56	0.47	0.21***	2.70	0.47	0.22***
FamSS	−0.86	0.19	−0.18***	−0.91	0.20	−0.19***
CNexFamSS				0.49	0.28	0.07
Child	R^2 = .18			R^2 = .19
Neglect	F(4, 578) = 32.20***			F(5, 577) = 26.46***
				F$\mathrm{\Delta }$(1, 577) = 3.08, ns
Age	−0.13	0.03	−0.18***	−0.13	0.03	−0.18***
Educ	−0.55	0.13	−0.17***	−0.55	0.13	−0.17***
HCh	0.61	0.27	0.09*	0.60	0.27	0.09*
FamSS	−1.05	0.19	−0.22***	−1.05	0.19	−0.22***
HChxFamSS				−0.03	0.16	−0.01
Household	R^2 = .15			R^2 = .15
Challenge	F(4, 575) = 24.86***			F(5, 574) = 19.86***
				F$\mathrm{\Delta }$(1, 574) = 0.05, ns

Notes: Presented values are from Steps 1 and 2 of the regression models with perceived family support as a moderator of the associations between childhood adversities and adult anger. Educ = Education, CAb = Child Abuse, FamSS = Family Social Support, CNe = Child Neglect, HCh = Household Challenge. *p < .05, ***p < .001.

3.4.2. Child neglect

The model including age, education, child neglect, and perceived family support was significant and explained 19% of the variance in anger (Table 4). Main effects matched those in the child abuse model. The interaction effect between child neglect and perceived family support was not significant.

3.4.3. Household challenge

The model including age, education, household challenge, and perceived family support was significant and explained 15% of the variance in anger (Table 4). Main effects were the same as the child abuse model. There was not a significant interaction between household challenge and perceived family support.

4. Discussion

Veterans in NI reported a high number of ACEs (McGlinchey & Armour, 2023; McLafferty et al., 2015, 2021), and therefore, based on extant research, may be at an increased risk of developing problematic anger, contributing to higher suicide attempts and behavioural and functional health issues (Adler et al., 2020; Naifeh et al., 2021; Varker et al., 2022). The current study examined associations between ACEs and anger in greater detail and found that experiences of child abuse and neglect were most strongly associated with adult anger, in comparison to household challenge. Further, results highlight the protective impact of perceived social support, particularly from friends, on the association between child adversity and adult anger. However, perceived friend support was only protective for those that had experienced child abuse and household challenge, not child neglect. For veterans who were married or living with their partner, perceived partner support was also protective for those who had experienced child abuse, but no other type of childhood adversity.

4.1. Childhood adversity, perceived social support, and anger

Of the three categories of childhood adversity assessed, including child abuse, neglect, and household challenge, child neglect was the strongest predictor of adult anger in all regression models. The association between childhood neglect and anger was unmoderated by any type of perceived social support. These results are consistent with studies which have found child neglect has a stronger effect on adult depression (Merrick et al., 2017) and serious suicide attempts (De Araújo & Lara, 2016; Merrick et al., 2017) than child physical or sexual abuse. Further, negative perceived social support, including neglect, has been directly linked to greater PTSD symptom severity in military veteran men (Blais & Zalta, 2023). Experience of child neglect, rather than child physical or sexual abuse or household challenge, may be particularly salient for the development of adult anger in military veteran men. Elevated rates of anger have also been extensively linked to suicide risk, particularly for adults who have experienced multiple child adversities (Charak et al., 2016). Veterans who report elevated rates of anger and child neglect may therefore be a particularly at-risk population for mental ill health and suicide.

Experience of child abuse, on the other hand, was most impacted by the protective aspects of perceived support, both from friends and a long-term romantic partner. A prior study conducted with female adult survivors of abuse found that perceived family and friend support, but not partner support, moderated the association between child physical abuse and posttraumatic stress symptoms (Wilson & Scarpa, 2014). For women who reported high perceived family or friend support, there was no association between child physical abuse and posttraumatic stress. Male veterans tend to report high rates of child physical abuse (Koola et al., 2013; Lapp et al., 2005). In the NI context, child physical abuse was the highest reported ACE, with 32.8% of veterans reporting experiences of child physical abuse (McLafferty et al., 2021).

Our results support prior work that has found a protective effect of perceived friend support with women (Wilson & Scarpa, 2014), and differ in that we did not find a protective effect of family support. In the current study, perceived partner support, but not family support, was protective against the development of anger for male veterans who experienced child abuse. However, this was true only for veterans who reported the highest perceived partner support possible using the MSPSS. This suggests there is a gender difference in how perceived social support, particularly sources of perceived support, impacts associations between child abuse and mental health outcomes. Perceived friend support may be a social resource for adaptive processing of child abuse for both males and females. Further, perceived family, friend, and partner support represent very different relationships and durations of support, or perceived lack of support in the case of family as the source of abuse.

Although perceived friend and partner support had mitigating effects on the association between child adversity and anger, perceived family support did not play the same role. Perceived family support was generally adaptive, in that it had a direct negative effect on anger. However, for adults that reported higher rates of child abuse or neglect or household challenge, perceived family support was not protective against adult anger. This particular result did not support our general hypothesis that perceived support would moderate the association between child adversity and anger. Considering military populations report higher rates of ACEs (Katon et al., 2015; McLafferty et al., 2021), many may join the military as a way of escaping a chaotic or abusive home environment. Therefore, for some men whose family was the source of abuse, family support may not moderate associations among childhood adversities and anger as it is not welcome due to mistrust and vigilance of threat (Miller et al., 2011). Men who do not welcome family support due to experiences of child adversity may therefore place greater value in friend or partner support from a young age.

4.2. Perceived friend support and military veterans

For veterans who perceived high social support from their friends, there was not a significant association between child abuse or household challenge and anger. However, when veterans perceived low social support from friends, associations between child adversity and anger were positive. These results imply that perceived support from friends is particularly important for veterans who have experienced different types of adversity as a child, and that this perceived support may lead to better mental health outcomes after transitioning out of the military. A meta-analysis by Zalta et al. (2021) found that social support was particularly important for veterans exposed to war or combat. Time since trauma did not impact the association between social support and PTSD, suggesting that later social support can have beneficial impacts on PTSD severity long after a traumatic event. Hinojosa and Hinojosa (2011) found military friendships to be a key factor in successful reintegration post-deployment for veterans. The importance of friendships for military service members may stem from the tight-knit bonds that often form as part of military units leading to greater emotional relatedness to unit peers (Kaplan & Rosenmann, 2014).

However, in a recent meta-analysis by Blais et al. (2021) specifically focusing on PTSD and social support in military veterans, perceived non-military support had a stronger effect on PTSD symptoms than perceived military support. Further, social support received outside of, rather than during, deployment was more beneficial in terms of PTSD symptoms, although this lost significance when methodological covariates were considered. This suggests that while military friendships are beneficial, perceived support outside of the military may be more important for issues such as PTSD and anger. Although our measure of perceived friend support did not distinguish between military and non-military peer support, our results also highlight the beneficial impact of general perceived peer support on adult anger for military veterans.

These results have applied implications. Promotion of male community support groups, such as Men’s Sheds, to service members transitioning out of the military may minimise potential risk for problematic anger and mental and physical health problems. Military organisations might also endeavour to include programmes which help veterans to overcome barriers to support seeking and establish non-military friendships and peer support. This might be particularly impactful for veterans in NI, who are more likely to experience stigma because of their British military status due to The Troubles, a 30 year long civil conflict in NI from the 1960s-1990s in which the British military had a significant role (Operation BANNER, 1969-2007). This would also have particular benefit for military men with ACEs in which their family was the perpetrator, as they are susceptible to compounding risks for anger and consequences such as suicide, due to experiences of childhood adversities, particularly neglect, and a lack of a key source of familial support.

4.3. Limitations and future directions

The current study is cross-sectional in design, meaning that we cannot determine true direction of effects, only that there are significant associations. However, there is an inherent longitudinal aspect and temporal ordering as we assessed retrospective adversity experienced in childhood and current reports of perceived social support and anger. The study is also limited by the use of self-report measures. Retrospective self-reports, particularly of adverse experiences in childhood, may be biased, and adults may not want to report certain experiences due to stigma. A review on retrospective reporting of ACEs found that adults may be underreporting the amount or type, except in the case of severe abuse or neglect, of childhood adversities (Hardt & Rutter, 2004). Furthermore, the measurement of ACEs in the study relied on a measure which equates all ten ACEs as being equal to one other, so there may be important and nuanced distinctions among adverse experiences that are not captured in the current study (Lacey & Minnis, 2020). We are also missing information regarding the intensity, frequency, and duration that these childhood adversities occurred, which may have important implications for outcomes such as anger. Future work might employ a more comprehensive assessment of ACEs to further understand these impacts in military veterans. Reporting of anger may also be influenced by use of self-reports, rather than clinician ratings.

The design of the survey used in this study allowed participants to skip questions they did not want to answer, and so there is a large amount of missing data. Although this encouraged veterans to participate, it meant that more than half of the cases were deleted in the current study due to incomplete data across core measures of interest. Further, this sample may not be representative of the military veteran population residing in NI, although it is substantially similar to the Ministry of Defense (2019) report regarding core sociodemographics that make up the veteran population in the UK. Due to lasting effects of The Troubles, many veterans are still wary of revealing their veteran status and may have been/be reluctant to participate in research. In the current study, we also found that excluded veterans who were missing 100% of data on anger, ACEs, or social support had significantly lower educational attainment than included veterans who responded to at least some items on the measures of interest. Therefore, the length and comprehensiveness of the survey may have caused excessive response burden (Rolstad et al., 2011) on veterans with lower levels of educational attainment. Future research might consider including less or shorter measures to ensure greater participant response.

Future research may benefit from a larger-scale population sample of veterans in NI, and the use of clinician ratings for anger or other components of mental health, which would also decrease participant response burden. The current study highlighted the protective role of perceived friendship in dealing with childhood adversity and adult anger. Future work may wish to examine the role of perceived social support and common comorbidities of anger, such as PTSD and mood disorders, to tackle the concerning suicide rates in military veteran populations.

5. Conclusions

Perceived friend support was particularly beneficial for veteran men regarding responding adaptively to child abuse and household challenge. Results concerning child neglect suggest that this type of childhood adversity is concerning due to its strong association with anger, unchanged by veterans’ perception of support they receive from their family, friends, or partner. Increased screening of childhood adversities during recruitment can inform preventative, rather than curative, programmes and treatment. This could reduce problematic anger in military servicemen, and, in turn, serious mental ill health and/or suicidality. To provide effective resources to service members transitioning out of the military, government and non-profit organisations may wish to create or promote integrated military and non-military community support groups for men.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Participants did not consent to their data being made publicly available. The raw data corresponding to the paper may be made available upon reasonable request from the Principal Investigator (Cherie Armour) in conjunction with an appropriate data sharing agreement.

Additional information

Funding

The authors would like to thank the Forces in Mind Trust [award number FiMT15/0624UL/ NIMH] for funding the Northern Ireland Veterans’ Health and Wellbeing Study. The funders had no role in the study design or publication of results.