Enablers and barriers to military veterans seeking help for mental health and alcohol difficulties: A systematic review of the quantitative evidence

Abstract

Objective

Research exploring the enablers and barriers that exist for military veterans seeking to address their poor mental health has produced ambiguous results. To identify the enablers and barriers correctly, this study systematically reviews the literature, including research that included alcohol and had a clearly defined veteran population.

Methods

Six databases were searched. Inclusion criteria specified that empirical studies related to veterans that had ceased military service and were seeking help for poor mental health and/or alcohol difficulties. Critical Appraisal Skills Programme and AXIS appraisal tools were used to assess quality and bias. A narrative synthesis approach was adopted for analysis. From 2044 studies screened, 12 were included featuring 5501 participants.

Results

Forty-four enablers and barriers were identified, with thirty-two being statistically significant. Post-traumatic stress disorder had the greatest number of enabler/barrier endorsements to veterans seeking help. Depression, anxiety, experience and attitudes also acted as enablers/barriers. Most studies were of fair methodological quality. Limitations included that samples were skewed towards US army veterans. Little research exists concerning those that have ceased military service.

Conclusions

Veteran help-seeking is likely enabled by poor mental health symptomology and comorbidity, which suggests veterans reach a crisis point before they seek help. Further research on alcohol misuse and attitude formation is required. The field would also benefit from alternative study designs including qualitative studies with non-US participants.

Introduction

Existing veteran mental health (MH) literature suggests veterans have a greater vulnerability towards mental illness than the general population. When Rhead et al. ¹ compared UK veterans with the general population and found veterans were 30% more likely to suffer from common MH disorders. One explanation relates to elevated levels of career-specific trauma exposure encountered by veterans.^2,3 Veterans were nearly twice as likely to suffer post-traumatic stress disorder (PTSD) and alcohol problems when compared with the general population. ¹

Jones and Fear ⁴ reported that the military has a long history of considering alcohol as a ‘cure-all problems’ substance. Historically, alcohol was handed to soldiers to relieve the anxieties associated with difficult and traumatic combat situations. Over time, alcohol within the military became normalized and was consumed excessively for stress management, commemorations, celebrations, peer bonding, pain management, insomnia and relaxation. Normalized military-shaped beliefs and attitudes often cascade into veteran life; thus, alcohol may continue to be consumed excessively post-military service. ⁵ In their profile study of 600 help-seeking UK veterans, Murphy et al. ⁶ found that while 43% reported problematic alcohol use as their main MH difficulty, 93% suffered alcohol issues comorbidly with two to three other MH problems. These findings suggest excessive alcohol use is an issue for many veterans, in the context of other mental illnesses.

Particularly within military culture, help-seeking for poor MH and problematic alcohol use is often delayed or avoided. According to a review by Hom et al., ⁷ approximately only one-third of the armed forces community globally engages in help-seeking via health care or professional organizations. This is not to suggest that only one-third of this population needs to seek help but, rather, there may be many more who could benefit from help-seeking but who choose not to. In any event, global veteran help-seeking figures are unlikely to be entirely accurate, as much help-seeking research is conducted in the US where veterans have a different health care system to other countries (e.g. UK and Australian veterans often use channels that would not necessarily recognize or record them as veterans).

However, it is generally reported across the literature that help-seeking rates are relatively low for veterans. ⁸ Several reasons for this have been proposed. For one thing, aside from alcohol being (potentially) a barrier to adaptive help-seeking, veterans may also delay or avoid seeking help due to military ideology and veteran culture. Research indicates that help-seeking may be inhibited by veterans’ strong sense of pride and a fear of stigma.^9–12 An alternative explanation is that veterans have limited knowledge or negative attitudes towards MH. ¹³ A 2017 review by Kantor et al., ¹⁴ which included participants from the armed forces, found that limited MH knowledge was a barrier to help-seeking. The Health Belief Model ¹⁵ proposes that attitudes towards symptoms and treatment can enable or hinder help-seeking; therefore, it could also be that veterans have had negative treatment experiences on occasions when they did seek help, in turn reinforcing their negative attitudes towards care. Poor help-seeking experiences have been suggested as potential barriers to future help-seeking for veterans.^7,16 Johnson and Possemato ¹⁷ found that cognitive predictors, such as problem recognition and treatment beliefs, were linked to support service use over and above practical factors. Rafferty et al. ¹⁸ also found problem recognition was a key barrier to help-seeking.

While the existing literature does indicate that some veterans avoid seeking help, there are many veterans who do seek support. Murphy et al. ⁶ found that a main motivator for help-seeking in UK veterans was severity of poor MH symptomology, particularly PTSD. It was not necessarily the presence of PTSD symptoms that enabled help-seeking but rather the extent to which the symptoms impacted the veterans’ ability to function and quality of life. A further potentially salient enabler of help-seeking is positive social support. ¹⁹ When others within a positive support network recognize the signs of poor MH, they often offer a veteran the requisite encouragement and assistance (physical and emotional) to seek support. ²⁰

Previous systematic reviews have been conducted to explore enablers and barriers to veterans’ help-seeking (see online Supplement). But these have often produced inconsistent findings. One possible explanation for this could relate to mixing military-affiliated populations within the same review. For example, in the US the definition of a ‘veteran’ is often someone who has been operationally deployed, whereas in the UK it is someone who has ceased military employment. ²¹ Arguably, a deployed (yet still employed) veteran faces different barriers and enablers to a retired veteran. An employed veteran may suffer workplace stigma ²² as a barrier, whereas a retired veteran may experience support access barriers. ¹⁴ Randles and Finnegan’s review ²³ concluded that research based on the US-veteran definition lacks generalizability to other veteran populations, which, in turn, is a limitation for reviews that include both US and non-US veteran studies.

A further source of inconsistency relates to alcohol. There is a lack of agreement over whether alcohol is a barrier to help-seeking. Some reviews failed to include alcohol altogether,^23,24 despite the literature suggesting that MH and alcohol misuse are linked. A final critique relates to the appropriateness of mixing quantitative and qualitative methodologies within the same review. Several mixed-method reviews provided no details of study design or justification for mixing the methods, meaning it was difficult to assess the quality of the reviews.^25,26

To our knowledge, the current systematic review is the first to include alcohol within the context of help-seeking enablers/barriers for veterans who have ceased military employment.

The purpose of this study was to address two questions:

• (1) What are the barriers and enablers to help-seeking for veterans having ceased military employment, for those with MH and/or alcohol issues (comorbid or otherwise)?

• (2) How many enablers and barriers exist?

Methods

We followed the Centre for Reviews and Dissemination’s guidance regarding search strategy development, inclusion and exclusion criteria, quality assessment, data extraction and data synthesis. ²⁷ A systematic review protocol was registered on PROSPERO, reference CRD42021227097. ²⁸

Eligibility criteria

Studies were considered for inclusion if they met the following criteria:

• ⁃ Not currently serving personnel. The target population was defined as veterans consistent with the UK definition of a veteran. ²¹ Reserves were excluded, as they could technically still be classed as serving.

• ⁃ Featured or measured enablers and barriers to help-seeking contact or initial service use. Support services were either MH-specific and/or could assist MH indirectly.

• ⁃ Included measures of MH and alcohol to assess if these variables affected help-seeking.

• ⁃ Retrospective (‘would you consider help-seeking?’) or actual help-seeking behaviour (did/did not help-seek).

• ⁃ Measured correlates of actual or retrospective help-seeking behaviour.

• ⁃ Quantitative study designs.

Exclusion criteria included the following:

• ⁃ Non-veteran populations (an exception to this was where the general population was considered as a comparison group).

• ⁃ Long-term, repeated service use or attrition studies, as long-term service use is not considered comparable to the more initial stages of help-seeking.

• ⁃ Studies featuring general substances, which did not differentiate alcohol misuse.

• ⁃ Non-English studies not using primary data collection and non-peer reviewed papers.

• ⁃ Qualitative studies.

Selection process

CH electronically searched six databases on 21 December 2020 and again on 1 December 2021. For search inclusivity, no ‘search from’ date was included in the initial search, whereas the second search spanned 21 December 2020 to 1 December 2021. Databases were PsychINFO, Medline, Embase, CINAHL, SCOPUS and PILOTS. Keywords used within the literature search related to veterans, help-seeking, enablers, barriers, mental health and alcohol (see online Supplement for the full search strategy).

CH screened the sources identified. The screening process was piloted with a random sample of 20 studies. All titles and abstracts were screened, followed by full-paper eligibility checking. The reference lists of included articles were also screened for inclusion. A subset of 10% was cross-checked by PT at each stage.

Bespoke data extraction sheets were created, to extract information relating to study objectives, participants, study design, outcomes, results, discussion, generalizability, limitations, funding and ethics. Data extraction sheets were piloted to test suitability. CH extracted all relevant data from included studies, which were checked by PT.

The literature search identified 2044 results. After removing 154 duplicates, 1890 titles and abstracts were screened. A full paper review was conducted on 253 papers, and 12 articles were selected for inclusion (see Figure 1). Details of the 12 included studies^29–40 are summarized in Table 1.

Figure 1.

Flow diagram on study selection.

Table 1.

Characteristics of the 12 studies.

Study	Study type	N	Participant characteristic	Objective, nature of study, main enablers and barriers identified
Bravo, Witkiewitz, Kelley & Redman (2019) 29	Cross-sectional – survey data	163	US, recruited through the Combat Wounded Coalition, male (95.7%), white (74.2%), army (45.5%), marines (25.6%), served in Afghanistan and Iraq (78.5%), reported a service-related disability (93.9%) 50.9% PTSD 64.5% depression 67.5% anxiety 16.6% suicide ideation 36.8% probable alcohol misuse	Objective: Would online mindfulness help veterans cope with emotional discomfort associated with combat injuries? What were the indicators of engagement? Nature: Retrospective study. Enablers: Depression, anxiety, anger, prior or current VA use for mental health and rumination. Barriers: Low self-compassion
Browne, Chen, Hundt, Hudson, Grubbs, & Fortney, (2021) 30	Cohort study – followed a group of veterans to mental health intervention from the point of recruitment. Initial attendance within the intervention was measured as an outcome	265	US, recruited through the VA Telemedicine Outreach for PTSD study, male (89.91%) and female (10.19%) veterans, met the diagnosis criteria for PTSD, 63.77% white, 29.62% African American, Afghanistan and Iraq veterans (29.06%), other conflicts (70.94%), combat role (49.43%) 100.0% PTSD 78.9% depression 67.2% anxiety 77.7% probable alcohol misuse	Objective: Exploring rates and reasons associated with non-attendance in therapy for PTSD Nature: Actual behaviour study Enablers: None (barrier study) Barriers: Perceptions of therapy, embarrassment to discuss feelings, no transport, no childcare, schedule conflict and poor experience of treatment
Di Leone, Gradus, Giasson, Vogt, Street, & Resick, (2013) 31	Cross-sectional – survey data	1040	US, recruited through the VA, male and female (51%) veterans, completed at least one year of active service in Iraq and/or Afghanistan, and have departed active service. 63.2% army. Male 80.6% white, 10.95 African American, 8.5% Other 37.4% PTSD 13.2% depression 17.3% anxiety 2.7% probable alcohol misuse	Objective: What predicts whether veterans with a mental health need are receiving care? Nature: Actual behaviour and retrospective attitudes were measured Enablers: Income, ethnicity, sexual harassment (for women), warfare exposure, PTSD, anxiety, depression, and perceptions of services Barriers: None
Erbes, Westermeyer, Engdahl, & Johnsen (2007) 32	Cross-sectional – survey data	117	US, recruited through the VA, male (86%) and female, 91% army, completed survey six months after active service in Iraq, Afghanistan and other (24%). 12.0% PTSD 33.0% probable alcohol misuse	Objective: Evaluate levels of poor mental health outcomes in conjunction with mental health service use, and whether mental health outcomes predicted or correlated with mental health service use Nature: Actual behaviour study Enablers: PTSD and depression Barriers: None
Fox, Meyer, & Vogt (2015) 33	Cross-sectional – survey data	278	US, recruited from the Defence Manpower Data Centre roster, male and female (59%) veterans, 64% white, 19% African American, 65% army, Iraq and Afghanistan veterans. 43.0% PTSD 80.4% depression 35.0% probable alcohol misuse	Objective: Assess attitudes regarding mental health care and mental illness beliefs, together with the participants’ relationship with the VA mental health-specific service Nature: Retrospective and actual behaviour. Enablers: Perceptions of help-seeking, perceptions of mental health issues, entitlement to care, similarity to other VA users, and mental health comorbidity Barriers: None
Marshall, Jorm, Grayson, Dobson, & O’Toole (1997) 34	Cross-sectional – face-to-face structured interview, plus self-completion mail survey	641	Australia, recruited through the VA, restricted to male army Vietnam veterans Unreported % mental health outcomes	Objective: Did PTSD and other predictors enable help-seeking of mental health support? Nature: Retrospective and actual behaviour. Enablers: General mental health, physical health, personal problems, family relationships, marital dysfunction, and repatriation issues Barriers: Repatriation issues
Owens, Herrera & Whitesell (2009) 35	Cross-sectional – on-line survey	50	US, recruited veteran interest groups and adverts, female-only veterans, served in Iraq or Afghanistan, 74% were white, 8% African American, 61% army. 24.0% PTSD 100.0% depression 66.0% probable alcohol misuse	Objective: What were the enablers and barriers to mental health support for female veterans? Nature: Retrospective and actual behaviour. Enablers: Income, gender, abuse, PTSD, anxiety, depression, anger, eating disorder, substance abuse, marital dysfunction, seen as weak, embarrassed to help-seek, felt unwelcome, and career harm Barriers: Gender, waiting time, warfare exposure, seen as weak, embarrassed to help-seek, felt unwelcome, career harm, and experience
Ryan, McGrath, Creech Y Borsari (2015) 36	Cross-sectional – on-line survey	133	US, recruited through the VA, female only, Afghanistan or Iraq veterans, 80.9% white 22.1% PTSD 16.4% problematic alcohol use	Objective: The role of trauma and PTSD symptom predictors of mental health care engagement Nature: Actual behaviour measured Enabler: PTSD Barriers: None
Tsai, Mota, & Pietrzak, (2015) 37	Cross-sectional – on-line survey	1202	US, through the National Health and Resilience Veterans Study, users of the VA (829) and non-users, females and males, but mainly featured female use. VA female users were 56% white and 28% black. Female other service users were 67% white and 15% black. Men VA users were 69% white and 14.3% black. 77% of female VA users served in Iraq or Afghanistan, with 9.7% serving in Vietnam. 43% female other-service users served in Iraq or Afghanistan, with 36% serving in the Persian Gulf war. 23% of male VA users serve in Iraq or Afghanistan with 48% serving in Vietnam Unreported% mental health outcomes	Objective: To determine enablers and barriers to mental health service use, including VA-specific care Nature: Retrospective and actual behaviour. Enablers: Income, ethnicity, PTSD, depression, substance use (alcohol), physical health, combat veteran status, and served on a particular tour. Barriers: Cost, seen as weak
Vogt, Fox, & Di Leone (2014) 38	Cross-sectional – survey data	640	US, recruited from the Defence Manpower Data Centre roster, returned from deployment 2–4 years prior, served in Iraq and Afghanistan, women 56.7%, Caucasian (66.8%), army (50.7%) 19.5% PTSD 36.6% depression 15.6% probable alcohol misuse	Objective: Examine mental health beliefs and association with support service use, for poor mental health and substances Nature: Retrospective and actual behaviour. Enabler: Perceptions of mental health services Barriers: PTSD, depression, substance (alcohol), work relations, perceptions of help-seeking, and preference for self-care
Winn, Shealy, Kropp & Felkins-Dohm (2013) 39	Cohort design – followed a homeless group of veterans from exposure to an intervention, through to a treatment programme. The aim was to get the veterans to help-seek via a treatment programme	211	US, veterans recruited through the VA. All had been recommended to attend an addiction treatment programme. Male (95%) and female. 63% were white, 33% were African American, 27% were homeless or in some sort of supported housing 100.0% substance use disorders	Objective: Does house assistance improve access to substance use treatment? Nature: Actual behaviour Enablers: Housing, legal problems, family relationships, caseworker assistance, and experience Barriers: None
Woodhead, Ronal, Iversen, MacManus, Hotopf, Dean, McManus, Meltzer, Brugha, Jenkins, Wessely, & Fear (2011) 40	Cross sectional – survey data	761	UK. Data analysed from the Adult Psychiatric Morbidity (general household) Survey. Post-national service veterans. Non-veterans (504) were matched to veterans (257) by age and gender, to form a comparison group. 85.2% male 7.4% PTSD 34.9% any neurotic disorder 53.3% common mental disorder (includes alcohol and neurotic disorders) 60.2% suicidal thoughts 16.3% suicide attempts 17.7% self-harm 18.2% probable alcohol	Objective: Assessing predictors of poor mental health and mental health service use Nature: Actual behaviour Enabler: General mental health Barrier: Length of service

VA = Veteran Affairs, a government department. Veteran may be recruited through the VA (as the VA may hold veteran records), but not all veterans use VA services. VA affiliation does not necessarily equate to VA service use.

Due to study heterogeneity, articles were synthesized by narrative analysis. From this, a detailed table of enablers and barriers was created, listed in Table 2.

Table 2.

Enablers and barriers.

Study	Category/item	Endorsed as enabler and/or barrier	Statistically significant a	Effect size (OR) b
	Demographic
31, 35 and 37	Low income	E	31 and 37	2.06 31 –2.71 31
31 and 37	Ethnicity (minority race)	E	31 and 37	1.98 37
37	Employment	E	37	-
39	Housing	E	39	4.73 39
35	Gender – not enough for women	E and B	-	-
	Trauma type
31	Sexual harassment	E	31	-
31 and 35	Warfare exposure	E and B	31 and 35	1.01 35
35	Abuse	E	-	-
	Mental health condition
29, 31, 32, 35, 36, 37 and 38	PTSD	E and B	29, 31, 32, 35, 36, 37 and 38	1.03 31 –2.85 37
29, 31, and 35	Anxiety	E	29, 31	3.40 31
29, 31, 32, 35, 37, and 38	Depression	E and B	29, 31, 32, 37 and 38	0.88 38 –3.39 29
29 and 35	Anger	E	29	3.01 29
35	Eating disorder	E	-	-
35, 37, and 38	Substance misuse	E and B	37 and 38	0.41 37 –1.47 37
34 and 40	General mental health c	E	34 and 40	1.12 34
	Interpersonal factors
34 and 39	Family relationships	E	34 and 39	1.06 34
34 and 35	Marital dysfunction	E	34	1.06 34
38	Work relations	B	38	2.97 38
	Military factors
34	Repatriation issues	E and B	34	2.97 34
34 and 37	Combat veteran	E	34 and 37	1.09 34
37	Served in a particular tour (Iraq)	E	37	-
40	Length of service	B	40	-
	Practicalities
35	Waiting time	B	-	-
39	Caseworker assistance	E	39	4.50 39
37	Cost	B	37	0.76 37
30	Transport	B	-	-
30	Childcare	B	-	-
30	Schedule conflict	B	-	-
	Attitudes
33 and 38	Perceptions of help-seeking	E and B	33 and 38	0.86 33 –0.90 33
30, 31, and 38	Perception of services	E	31 and 38	1 × 10 31 –1.17 31
33 and 38	Perceptions of mental health issues	E and B	33 and 38	0.86 33 –1.01 33
33	Entitlement to care	E	33	1.14 33 –1.17 33
33	Similarity to other VA users	E	33	1.15 33
35 and 37	Concern over being seen as weak	E and B	-	0.61 37 –1.48 37
30 and 35	Embarrassed to help-seek or discuss issues	E and B	-	-
35	Feel unwelcome	E and B	-	-
35	Concerns over career harm	E and B	-	-
38	Preference for self-care	B	-	-
29	Tendency to ruminate	E	29	1.39 29 –1.55 29
29	Tendency to have low self-compassion	B	29	0.94 29
	Non-categorized (Es) and (Bs)
34 and 37	Physical health d	E	34 and 37	0.18 37 –1.61 34
34	Personal problems	E	34	1.06 34
39	Legal problems	E	39	2.37 39
29, 31, 35, and 39	Experience	E and B	29 and 39	0.96 29 –5.28 29

^aStatistical significance – only those studies that reported significance feature here.

^bORs – Odds ratios were included only for those studies that reported them. We list the smallest to largest ORs across those studies that report them.

^cGeneral mental health. The first study ³⁴ refers to general mental symptoms, depression, marital dysfunction, PTSD symptoms, PTSD lifetime diagnosis, PTSD avoidance, agoraphobia, social phobia, simple phobia and lifetime alcohol. The second study ⁴⁰ refers to general mental health as any/or mental health issue captured in the study – severe alcohol use, alcohol dependence, PTSD, drug dependence or any neurotic disorder.

^dClassified as high or low risk of having/developing physical health issues, based on answers collected via a standardized government health interview. The questions related to utilization of services relating to chronic illness, with risk factors linked to alcohol, smoking and exercise.

The quality of each study’s research method was assessed via two critical appraisal tools: the Critical Appraisal Skills Programme (CASP) for assessing cohort studies ⁴¹ and the AXIS ⁴² tool for assessing cross-sectional studies. The Equator website ⁴³ was also referred to regarding reporting transparency and quality of health research. A quality assessment matrix was created based on the CASP and AXIS indicators of quality, ⁴⁴ with each study awarded a score out of 25. CH assessed the overall quality of the research, with PT cross-checking a 10% subset.

Results

Study characteristics

Ten studies were cross-sectional,^{29,31–38,40} with the other two^30,39 being cohort studies. Only one study ²⁹ was solely retrospective regarding help-seeking.

Ten out of the twelve studies^{29–33,35–39} were from US, conducted with Iraq and Afghanistan veterans. Although these were US studies, the personnel in these studies nevertheless conformed to the UK definition of a veteran. Of the two remaining studies, one was Australian ³⁴ and the other United Kingdom. ⁴⁰ In total, the studies included 5501 participants, with 4099 (74.51%) participants being US veterans and 2014 (36.61%) identifying as serving in the army. Males accounted for 3329 (60.52%) of all participants, and Caucasians were represented by 2768 (50.32%). Not all studies reported each demographic described here. Six studies^{30–32,34,36,39} recruited participants via the Veteran Affairs (VA) administration, amounting to 2290 (41.63% of all participants). Across all studies, the proportions of participants that were reported as having likely or actual MH and alcohol difficulties were PTSD 7.4–100%, depression 13.2–100%, anxiety 17.3–67.5% and alcohol misuse 2.7–77.7%. One study ³⁹ reported 100% of the population were misusing substances, but it is unclear how many experienced alcohol misuse.

Enablers and barriers

All the studies, bar one, ³⁰ identified enablers. Seven studies^{29,30,33,35,37,38,40} identified barriers. See Table 1 for a full list of enablers and barriers.

The total number of individual enablers and barriers was 44. Thirteen barriers and enablers were regarded as both barriers and enablers, including PTSD and repatriation issues. Thirty-two barriers/enablers were statistically significant.

The most common MH conditions found to predict help-seeking were PTSD (most frequently occurring predictor), depression (strongest MH predictor, OR up to 3.39²⁹), anxiety and substance misuse:

• • PTSD (presumed or actual) significantly predicted help-seeking across seven studies.^{29,31,32,35–38} Two further studies^34,40 found ‘general’ MH was a significant enabler, with PTSD being captured within general MH measures. One study ³² reported that PTSD was the most significant predictor when controlling for all other correlating variables. Another study ²⁹ found PTSD symptom severity acted as a help-seeking enabler. However, one study ³⁸ suggested that when PTSD correlated with negative beliefs help-seeking reduced – therefore, PTSD was considered to act as a help-seeking barrier within this study.

• • Depression was found to be a help-seeking enabler or barrier across six studies.^{29,31,32,35,37,38} Five studies^{29,31,32,35,37} suggested that depression predicted help-seeking and acted as an enabler. One study ³⁸ found that depression acted as a barrier when correlated with negative beliefs.

• • Anxiety was found to be a help-seeking enabler across three studies.^29,31,35

• • Substance misuse predicted help-seeking across three studies.^35,37,38 In one study, ³⁵ general substance misuse predicted help-seeking, but alcohol specifically did not. Alcohol was a significant enabler in one study. ³⁷ Yet, alcohol was a significant barrier in another study ³⁸ when correlated with negative beliefs.

Experience had the strongest (enabling) effect on help-seeking, increasing the odds of help-seeking by up to 5.28²⁹ times. Housing or caseworker assistance also had a large enabling effect (ORs 4.73 and ³⁹ 4.50 ³⁹ ). Low income^31,35,37 (enabler), repatriation ³⁴ (enabler/barrier) and work relationships ³⁸ (barrier) saw similar effect sizes of approximately three times on help-seeking. Regarding MH and alcohol outcomes, anxiety had the greatest enabler effect (OR 3.40 ³¹ ), followed by depression (OR up to 3.39 ²⁹ ), PTSD (up to OR 2.85²⁷), and substance misuse (up to OR 1.47 ³⁷ ).

Other regularly occurring individual enablers/barriers to help-seeking were experience of service use and perception of mental health services. Experience of service use was endorsed across four studies,^29,31,35,39 as both an enabler and barrier to help-seeking. Perception of mental health service use was endorsed across three studies^30,31,38 as a help-seeking enabler.

The final individual item, with three endorsements (as an enabler), was income.^31,35,37 The remaining 38 individual enablers and barriers were endorsed by one or two studies only.

We found we could aggregate forty one of the enablers and barriers into seven categories: demographic factors, trauma type, mental health condition, interpersonal factors, military factors, practicalities and attitudes. The remaining four uncategorized enablers and barriers were physical health, personal problems, legal problems and experience

The attitudes category contained the most individual enablers and barriers – twelve different help-seeking enablers and barriers were reported across seven studies.^{29,30,31,33,35,37,38} Within the category of attitudes, perception of services had the most enabler endorsements, with MH-specific rumination having the largest enabler effect (OR 1.55 ²⁹ ). Also, within a single category, attitudes had the most items that were endorsed as both help-seeking enablers and barriers (six items^{30,33,35,37,38}).

Quality and robustness of research

We assessed the methodological quality of most of the 12 studies as fair, with overall scores ranging from 11 to 19 out of 25. The details are given in the online Supplement.

All studies provided clear objectives and described appropriate study designs to investigate those objectives. All studies clearly defined their participant populations, but there were differences in the populations recruited (e.g. MH status and deployment location). All studies collected quite detailed demographic information, but this varied across the studies concerning what was collected and how it was measured (e.g. education, income, employment and ethnicity were measured differently). Four studies^33,35,36,38 asked what MH issue(s) had prompted participants to engage in help-seeking.

We turn our attention now to the 10 cross-sectional studies. Only one study provided sample size calculations. Likewise, only one study ³⁸ oversampled women to create even gender ratios. Six studies^{30,31,33,34,35,39} featured a sample that had entirely screened positive for MH issues, had received a positive MH diagnosis, or who were actively help-seeking for MH difficulties. Three studies^33,34,38 mentioned randomization, which apparently related to participant selection, yet no study confirmed how randomization of participation occurred.

All 10 cross-sectional studies used questionnaires that were distributed in various ways (mail-back, web-based and face-to-face). The questionnaires utilized varied, totalling 41 different scales (see online Supplement for details). Most scales were valid and reliable. Two studies^33,38 used adapted scales and reported alpha scores to demonstrate scale reliability. Most studies^{29,31,33,35,36,38,40} reported cutoffs for all scales used where relevant (e.g. clinically relevant levels of MH condition), whereas some^32,34,37 reported no cutoffs or only for some of the scales. Several studies^32,34,37,40 had omitted some details of scale/scoring administration; thus, it was not possible to assess that the measurement instruments had been administered correctly.

All 10 cross-sectional studies, apart from two,^29,30 provided overall response rates together with reasons for dropouts. Despite some studies^29,32,35,36 having low response rate sample sizes, the sample size was deemed to be acceptable by the authors of those studies. Sample sizes also seemed comparable with similarly published studies. Three studies^31,33,38 statistically compared responders with non-responders, to check for response bias. All studies, aside from one, ³⁰ reported statistical significance; one study ³⁰ described its findings in relation to frequencies rather than statistical analysis. Three studies^30,32,35 did not report confidence intervals or odds ratios for the analysis conducted. One study ³⁷ set the confidence interval at 99% to improve the robustness of findings.

Regarding the two cohort studies,^30,39 both included an intervention with separate conditions. However, it is unclear whether the studies had a within-participant or between-participant design. The studies inferred that participants could partake in all conditions, but this was not clearly reported. Therefore, it is unclear whether any randomization of conditions occurred.

Across all 12 studies, there was an inconsistency in the reporting of effect sizes (ORs) linked to enablers/barriers. ORs were only reported within studies^{29,31,33,34,36,37,38,40} (see Table 1 for the dispersion of reported effect sizes).

Discussion

This is the first review to identify enablers/barriers to help-seeking for veterans who have completely ceased military employment, in the context of alcohol as well as other MH difficulties. The inclusion of alcohol as a potential barrier/enabler is important as it often appears with other MH issues. Also, the specific definition of a veteran adopted in this review is salient as the term veteran differs globally, ²¹ which has implications for what is considered an help-seeking barrier/enabler. Forty-four different enablers and barriers were identified, with 13 more enablers being recognized than barriers. While this is contrary to Kantor et al.’s ¹⁴ review findings, who reported that more help-seeking barriers than enablers exist for veterans, our result is unsurprising as the focus of help-seeking research has shifted towards enablers over recent years.

Regarding individual enablers and barriers predicting help-seeking, PTSD was the most frequently endorsed enabler across not only MH conditions but also across all studies reviewed. This is to be expected, as research is often driven by the perspective that military veterans and PTSD difficulties are synonymous, ⁴⁵ which has created a PTSD-focused agenda within veteran research. Existing research shows that it is likely functional impairment caused by symptom severity (frequently associated with PTSD specifically) that leads to veteran help-seeking.^6,46 In this review, a veteran was up to three times more likely to engage in help-seeking if they had PTSD. While separate MH conditions were noted as individual barriers/enablers to help-seeking (and were mainly found to enable help-seeking – e.g. anxiety increased the likelihood of help-seeking by up to 3.5 times and depression up to 3.4 times), many studies found MH difficulties to be comorbid. This finding supports the view that veteran MH can be complex,^3,6 and it is often comorbidity complexity with symptom severity that drives help-seeking. ³⁹

Substance misuse, as an MH difficulty, was identified as impacting help-seeking; two studies found alcohol to be a significant enabler or barrier.^37,38 However, where alcohol was found to be a barrier it was correlated with negative beliefs or a negative disposition. ³⁸ Therefore, if beliefs were positive or negative beliefs were not included, alcohol would likely not be a barrier. One study ³⁷ found that alcohol had a stronger predictive effect on help-seeking via the VA for male veterans than females, even when controlling for other covariates. It was suggested within this study that male veteran VA users were more likely to screen positive for a lifelong alcohol use disorder than females. Incidentally, aside from study, ³⁹ it was difficult to distinguish whether participants were help-seeking for alcohol misuse specifically or other MH conditions, as few studies^30,35,39 reported which difficulties were being sought help for.

We now look further afield, at studies that were not included in our review because they did not match our selection criteria (e.g. they were qualitative papers or used a non-UK definition of a veteran). Our review has produced a salient finding to suggest that alcohol is not necessarily a main enabler or barrier to veteran help-seeking, a finding which aligns with the literature that suggests alcohol is not found to act as a help-seeking barrier.^47-49 A possible explanation could be that excessive alcohol use is associated with less functional impairment than, for example, PTSD symptoms, leading to a reduced perceived need for help. ⁴⁸ Both Gribble et al. ⁵⁰ and Hines et al. ⁵¹ found that where veterans may require support for problematic alcohol use specifically they may be less likely to help-seeking, due to a reduced sense of perceived need or problem recognition. A study that was included in our review, Erbes et al., ³² observed that heavy alcohol use could be associated with poorer quality of life in the domain of emotional dysregulation (and people may consume alcohol to manage the dysregulation). People with PTSD, in contrast, may suffer poorer quality of life and functional impairment in many other domains, thus prompting earlier help-seeking.

Other, more cognitive, factors were found to act as barriers and enablers to help-seeking. Experience of MH services (which in turn shapes feelings, thoughts and behaviours towards MH support services) had a high number of endorsements as enablers and barriers to help-seeking and the largest effect size. This finding was anticipated, as service experience has been found to predict help-seeking or future service use within the veteran community. For instance, several qualitative studies have investigated the process of veteran help-seeking and suggest that help-seeking is a journey that is motivated by mixed experiences.^20,52–54 However, help-seeking is not necessarily linear, as poor help-seeking experiences (creating poor feelings, thoughts and behaviours towards service use) can cause veterans to stall or regress with help-seeking (i.e. poor experiences become help-seeking barriers). ¹⁸

Another cognitive factor relates to perceptions of services. Perceptions of services received three endorsements^30,31,38 as an enabler to help-seeking, implying that positive perceptions of services are more likely to lead to help-seeking behaviour. This finding was mirrored by Johnson and Possemato’s ¹⁷ veteran cohort study of MH service use and attrition, which reported that treatment beliefs were a major enabler/barrier to service use. It could be inferred that positive perceptions of services are likely to be linked to positive service use experience, which impacts on the help-seeking journey and improves help-seeking motivation. ¹⁸ Positive perceptions also undoubtedly feed into positive health beliefs, causing positive health and treatment beliefs to act as a help-seeking enabler. Positive health and treatment belief leading to help-seeking is a core principle of the Mental Health Belief Model. ¹⁵

Perception of services is an example of ‘attitudes’ that may act as a barrier/enabler to help-seeking. Across the studies, a total of 12 separate enabler/barrier attitudes were identified, suggesting that attitudes, collectively, may act as a major barrier/enabler to help-seeking. This finding is unsurprising, as Johnson and Possemato’s ¹⁷ study relating to veteran support service help-seeking and attrition found that cognitive enablers and barriers had a greater impact than other, more practical, barriers. Johnson and Possemato found the strongest predictors of help-seeking were ‘attitude to the problem’ and ‘attitude towards mental health care’.

Further barrier/enabler attitudes identified through this review were a preference for self-care, being seen as weak, embarrassment and career harm. These enablers and barriers could be classified as military-shaped, stigmatized attitudes. Yet, several other non-stigmatized attitudes were also found to impact help-seeking in this review. Therefore, the view that it is fundamentally stigma as an attitude that impacts help-seeking is unsupported.^10,16 Stevelink et al. ⁵⁵ found in their study of help-seeking within the armed forces community that the third most cited reason for help-seeking was the realization they could not self-manage their issues, despite holding pro self-care beliefs. It is likely that symptoms had become so unmanageable that help-seeking was enacted despite holding stigmatized beliefs. Help-seeking could also be encouraged or arranged by the veteran’s social circle, while the veteran maintains their stigmatized beliefs. Johnson and Possemato ¹⁷ proposed that stigma can act as a barrier, but it is perhaps not the strongest barrier, as help-seeking occurs even in the presence of holding stigmatized beliefs and attitudes.

Surprisingly, social support was not found to be a main category in this review. Three studies^29,31,34 included scales that could be classified as social support measures. The first ²⁹ included a friendship scale, but found that friendship did not particularly support help-seeking and did not discuss this finding within the paper. The second ³¹ included a social support scale. However, that study was barrier-focused, and as social support is generally considered an enabler, it was not identified in the findings. The third ³⁴ included measures of repatriation, family and marital relationships. That study found that repatriation issues acted as a barrier, whereas marital and family relations acted as help-seeking enablers. Repatriation into the community had a similar sized effect to PTSD, which indicates its salience. The limited social support endorsements identified do support social support themes elicited in the extant veteran help-seeking literature. Evidence of family and marital relations acting as an enabler has been found,^19,20 together with help-seeking being enabled by wider community support via local repatriation. ⁵⁶

The final enabler/barrier to help-seeking relates to income, which was endorsed as an enabler in three studies.^31,35,37 We were not surprised by this finding; given that 80% of studies in this review were conducted in the US, where health care predominantly comes at a cost (albeit at a lower cost if via the VA ¹² ). As such, veterans were nearly three times as likely to help-seeking if they had a lower income, suggesting that veterans probably used the VA as this was the most affordable option for them. The effect of low income on help-seeking via the VA was comparable in size to that of having likely PTSD; thus, it was a relatively important driver of help-seeking behaviour. Income would not necessarily be endorsed as an enabler in other countries (e.g. the UK or Australia), where veterans could be cared for through a nationally available (largely free) health care system. That is not to say income is not potentially a barrier to those with access to free services. Those that reside in rural areas may have access issues such as transport costs. Access to services may also be more difficult where some preferred services are privately funded. Whether or not income is a barrier or an enabler to help-seeking for veterans is likely dependent on the context health care is delivered and how it is accessed.

The main design limitation in the research we reviewed was its lack of diversity. Most of the studies were cross-sectional analyses using samples predominantly composed of US male, army veterans, who served during the same conflicts. However, it has to be noted that cross-sectional data collection is a pragmatic method of accessing specific populations, such as hard-to-reach populations, such as veterans. ⁵⁷ Furthermore, as the veteran population predominantly consists of ex-army males (the army being the largest of the armed forces), the risk of using mainly male, army participants will not likely reduce in the future.

A final issue with the studies was that approximately 42% of participants were directly recruited through the VA. Such veterans could already be help-seeking and be on the VA ‘system’. If the findings of help-seeking research are to be used to encourage or help non-help-seeking to come forward, it is less helpful to conduct research focused on those who already engaged in help-seeking.

Limitations

There are two main limitations with this review. First, it was difficult to distinguish what help-seeking was conceptualized as. There is a difference between veterans reaching out in help-seeking over a defined period (e.g. sessions or weeks) versus longer term support use. We made a differentiation between more initial stages of help-seeking versus longer term support engagement. But the overlap between the stages of initial help-seeking and service user engagement is somewhat blurred. Therefore, two studies^30,39 in particular were included that could be viewed as straying from initial help-seeking.

Second, all the studies were in the English language. As such, our review could be viewed as skewed towards the experiences of English-speaking veterans in the West.

Conclusions

Collectively, it seems that veterans are reaching a mental crisis point before engaging in help-seeking. Although the findings relating to alcohol remain unclear, we can say that where veterans are choosing to manage symptom severity and functional impairment with alcohol, it is likely they are attempting to avoid some sort of crisis point. They may be unaware that they are potentially creating a comorbid MH issue, which will likely lead to worse outcomes ultimately.

While this review found PTSD was the main predictor of help-seeking, a broad range of other more cognitive enablers and barriers were elicited that are potentially modifiable, to either aid help-seeking or to potentially avoid a crisis point being reached. Practical barriers – mainly associated with access difficulties linked to cost, availability and location – could be addressed by support provider organizations. Attitudinal barriers/enablers could be improved by improving MH literacy and beliefs surrounding support. Such interventions need to be directed at both veterans and support service stakeholders, in order that stakeholders are fully invested in improving service user experience for veterans.^18,58

To design effective interventions, it is essential to understand more fully the underlying help-seeking processes specific to veterans classified as having ceased military employment. It would be beneficial to investigate such processes qualitatively, as context and culture may influence attitude formation. ⁵⁹ Aside from clarifying what veterans are help-seeking for, a qualitative approach may elicit clearer findings on the role alcohol plays.

ORCID iD

Catherine Hitch https://orcid.org/0000-0003-4685-8910