Association between modifiable social determinants and mental health among post-9/11 Veterans: A systematic review

Abstract

Introduction:

As U.S. Veterans reintegrate from active duty to civilian life, many are at risk for negative modifiable social determinants of health. The prevalence of mental health conditions among Veterans is also high. Awareness of the associations between these two factors is growing. This systematic review provides a comprehensive analysis of the current state of knowledge of the associations between modifiable social determinants and mental health among U.S. Veterans.

Methods:

The authors systematically searched four databases and identified 28 articles representing 25 unique studies that met inclusion criteria. Findings from the studies were extracted and synthesized on the basis of modifiable social determinants. Study quality and risk of bias were assessed using the Methodological Quality Questionnaire.

Results:

The studies identified in the systematic review examined three modifiable social determinants of health: 1) housing stability, 2) employment and finances, and 3) social support. Although the lack of validity for measures of housing stability, employment, and finances compromised study quality, the overall evidence suggests that Veterans with access to supportive social determinants had better mental health status. Evidence was particularly robust for the association between strong social support and lower symptoms of posttraumatic stress disorder.

Discussion:

Current evidence suggests the need to consider modifiable social determinants of health when designing mental health interventions. However, more research encompassing a wider range of modifiable social determinants such as food security, education, and transportation and using comprehensive methods and validated instruments is needed. Future research also needs to intentionally include Veterans from diverse racial-ethnic groups.

RÉSUMÉ

Introduction :

De nombreux(ses) vétéran(e)s américain(e)s qui réintègrent la vie civile après leur service actif sont vulnérables à des déterminants sociaux de la santé modifiables négatifs. La prévalence de troubles de santé mentale est également élevée chez les vétéran(e)s. La sensibilisation aux liens entre ces deux facteurs est également en croissance. La présente analyse systématique détaillée expose l’état actuel des connaissances sur les associations entre les déterminants sociaux modifiables et la santé mentale chez les vétéran(e)s américain(e)s.

Méthodologie :

Les chercheurs ont procédé à la fouille systématique de quatre bases de données — MEDLINE Ovid, Embase, PsycINFO et la Cumulative Index to Nursing and Allied Health Literature — et en ont extrait 28 articles représentant 25 études uniques qui respectaient les critères d’inclusion de l’analyse systématique. Ils ont relevé et synthétisé les observations des études au sujet des déterminants sociaux modifiables. Ils ont également utilisé le questionnaire sur la qualité méthodologique pour évaluer la qualité et les risques de biais des études.

Résultats :

Les études retenues dans l’analyse systématique évaluaient trois déterminants sociaux de la santé modifiables : 1) la stabilité du logement, 2) l’emploi et les finances et 3) le soutien social. Même si les mesures sur la stabilité du logement et sur l’emploi et les finances n’étaient pas valides, ce qui compromettait la qualité des études, l’ensemble des données probantes laissait entendre que les vétéran(e)s qui avaient accès à des déterminants sociaux favorables avaient une meilleure santé mentale. Les données probantes étaient particulièrement vigoureuses à l’égard de l’association entre un soutien social solide et des symptômes plus faibles d’état de stress post-traumatique.

Discussion :

Selon les données probantes actuelles, il est nécessaire de tenir compte des déterminants sociaux de la santé modifiables pour concevoir des interventions en santé mentale. Cependant, il faudra réaliser plus de recherches sur un plus vaste éventail de déterminants sociaux modifiables tels que la sécurité alimentaire, l’éducation et le transport, et recourir à des méthodologies détaillées et à des instruments validés. De futures recherche devront également inclure, à dessein, des vétéran(e)s de divers groupes raciaux et ethniques.

Mots-clés :

besoins sociaux, déterminants de la santé, emploi, ÉSPT, état de stress post-traumatique, Forces armées américaines, santé mentale, soutien social, stabilité du logement, suicide, trouble dépressif majeur, vétéran(e)s

LAY SUMMARY

Veterans who served post-9/11 face many challenges as they reintegrate into civilian life. Some of these challenges include securing stable housing, adequate food, employment, and social support, all of which are examples of social determinants of health. Veterans are also at risk for mental health conditions. Thus, this systematic review examined published articles to evaluate what is known about the relationship between social determinants and mental health among U.S. Veterans who served post-9/11. Using four large databases, the authors found 28 articles representing 25 unique studies. The identified studies reported on three social determinants of health: 1) housing stability, 2) employment and finances, and 3) social support. Robust evidence indicates that Veterans with strong social support had better mental health. The evidence for other social determinants of health was either weak or lacking. Studies needed stronger methods to measure housing and employment and finances, as well as more robust statistical analysis. In addition, the majority of U.S. Veterans who participated in the studies were non-Hispanic white men. More research on a wider range of social determinants — such as food security, education, and transportation — that uses stronger study methods is needed. Future research also needs to intentionally include Veterans from diverse racial and ethnic groups.

INTRODUCTION

U.S. Veterans who served post-9/11 face many challenges as they reintegrate into civilian life.^1–3 As they separate from active duty, Veterans leave a highly structured social environment, consistent income, and housing assistance.³ For many Veterans, this loss can result in a struggle for basic needs, such as food security and housing stability.^4–8 Food security and housing stability are two examples of social determinants of health. Social determinants refer to the conditions in which people live, learn, work, and play that affect a wide range of health and quality-of-life risks and outcomes.⁹

The social determinants of health framework, as shared by the Centers for Disease Control and Prevention (CDC), consists of five key areas: 1) health care access and quality, 2) education access and quality, 3) social and community cohesion, 4) economic stability, and 5) neighborhood and built environment.⁹ Many of these determinants are modifiable (e.g., housing stability, employment), and others are fixed (e.g., gender, race and ethnicity). Efforts to improve modifiable determinants would support Veteran health and well-being, especially with respect to mental health.¹⁰

Veterans who served post-9/11 have a high prevalence of mental health conditions, such as posttraumatic stress disorder (PTSD), major depressive disorder (MDD), and generalized anxiety disorder (GAD).^11,12 Risk of suicide is also a major mental health concern facing U.S. Veterans.^13,14 The high prevalence of these mental health conditions and concerns may be attributed to stress experienced during both combat and non-combat experiences, such as military sexual trauma.^11,15 Stress associated with transitioning back to civilian life can also contribute to, or exacerbate, poor mental health,^16,17 especially for Veterans who lack access to basic social determinants such as housing stability, employment or finances, and social support.^7,18,19

There is growing awareness of the need to consider the role of social determinants in mental health and mental health interventions among Veterans reintegrating into civilian life.^20,21 However, to the authors’ knowledge, there has been no comprehensive review of the associations between modifiable social determinants and mental health among Veterans to determine what is known and what knowledge gaps remain. Thus, this systematic review evaluated the current state of knowledge about the associations between modifiable social determinants and mental health among U.S. Veterans who served post-9/11. As this cohort of Veterans transitions out of active duty, they may be at risk for poor social determinants of health.²² In addition, post-9/11 Veterans served in the longest U.S. war and may thus be at increased risk of mental health concerns, along with transition-related stressors.^23,24 The knowledge accumulated from this review can be used to identify directions for future research and assist health care providers and policy makers to further identify and support the needs of Veterans.

METHODS

This systematic review followed guidelines from the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement.²⁵ Eligibility criteria for inclusion of studies in the systematic review were as follows: 1) full manuscript availability, 2) published in the English language, 3) based on empirical data published in a peer-reviewed journal, 4) published no earlier than 2001, and 5) findings specific to post-9/11 Veterans who served in the U.S. Armed Forces. Studies in which findings specific to post-9/11 Veterans could not be parsed out were excluded. All studies needed to examine an association between at least one modifiable social determinant of health (e.g., employment, food security, housing, social support) and at least one mental health condition or issue (e.g., MDD, GAD, PTSD), suicidal ideation, or both.

A medical librarian (AS) conducted the database search on Jun. 21, 2021. The databases searched were MEDLINE Ovid, Embase, PsycINFO, and Cumulative Index to Nursing and Allied Health Literature. The search strategy was first created in MEDLINE Ovid and then translated to the other databases, using both controlled vocabulary and keyword or phrase equivalents for the following three components: 1) post-9/11 Veterans, 2) modifiable social determinants, and 3) mental health. Search terms for modifiable social determinants of health included general terms such as social need and social determinants. More specific social determinants included housing, food security, utilities (e.g., electricity, water, heat, natural gas), social support, pollution exposure, air quality, employment, and financial stability. Search terms for mental health included general terms such as mental health, well-being, and psychological disorders. Specific mental health issues included stress, PTSD, depression, anxiety, substance use, and suicidal ideation. The complete MEDLINE Ovid search strategy is shared in the Appendix. In addition, a manual review of the reference lists of studies included in the systematic review was performed.

A flow diagram of the literature search and selection process is presented in Figure 1. After duplicate articles were removed, the initial search identified 1,102 articles. Three authors (NK, LS, and SR) reviewed the articles’ titles for topic relevance. They then reviewed the abstracts for the inclusion criteria using a two-pass system. At least two authors reviewed each article. Discrepancies were discussed until consensus was reached.

Figure 1.

Flow diagram of the literature search

The database search was updated on Aug. 29, 2022, to identify any new articles that were published during the manuscript preparation phase. After removing duplicates, an additional 90 new articles were identified. Of these, three abstracts were reviewed and ultimately two more articles were found that met inclusion criteria. The systematic review thus consisted of 28 articles. However, 4 articles originated from the same data set and thus were collapsed into 1 study for the purpose of data synthesis.^26–29 The final review encompasses 25 unique studies.

Once the authors identified the full-text articles that met the inclusion criteria, two authors (NK and SR) extracted data for synthesis. They reviewed each article for study design, sample sources, sample characteristics, study geographic locations and time period, types of social determinants and mental health conditions examined, operationalization of these variables, data analysis methods, and findings on the associations that fit the aims of the systematic review. This information was entered into an evidence table in Microsoft Excel, with each row representing a single article and each column representing a listed study characteristic. Using the evidence table, lead author NK and co-author SK synthesized the data across each study to look for patterns, similarities, and differences.

The authors did not perform a meta-analysis because the studies used an array of measures, variables, research designs, and statistical tests that would not support a quantitative synthesis of the literature. However, the authors share statistical outcomes from the studies (e.g., odds ratios and confidence intervals) when available to provide readers with a sense of the magnitude of the associations identified.

The authors used the Methodological Quality Questionnaire (MQQ) rating scale to assess bias and study rigor.³⁰ The questionnaire is used to assess nine study criteria: 1) theoretical or conceptual definition of the focal variables or constructs, 2) operational definition of the focal variables or construct, 3) research design, 4) sampling design, 5) sample characteristics, 6) reliability and validity evidence in quantitative studies and trustworthiness, credibility, and dependability in qualitative studies, 7) data analysis, 8) implications for topic-specific practice, and 9) implications for topic-specific policy. Each criterion was first scored dichotomously on its presence in the study (e.g., present = 1 point, not present = 0 points). If present, then the criterion was scored on its quality, with 0 points indicating poor quality and 2 points indicating good quality. These points were summed so each criterion had a score of 0 (lowest score), 1, or 3 (maximum score). The authors then summed scores for all the criteria to obtain a composite score that ranged between 0 and 27 points, with higher scores indicating higher study quality. Two authors (NK and SR) assessed each article’s quality independently using this questionnaire, and discrepancies were discussed with the team until a consensus was reached. The composite scores for the studies are provided in Tables 1, 2, and 3.

Table 1.

Results of individual studies for housing stability

Study	Study purpose	Study design and years	Sample size and characteristics	Mental health condition	Findings	MQQ score
Blakey et al. 35	Explored well-being among Veterans with PTSD and SUD and compared them with Veterans with PTSD only, SUD only, or neither disorder	Cross-sectional, 2009–2010	National sample, N = 1,102, male, 82.9%; white, 69.4%	PTSD SUD	Veterans with PTSD and SUD were more likely to report lifetime experiences of homelessness than the PTSD-only group. Weighted percentages of homelessness in the PTSD and SUD and PTSD-only groups were 15.30% and 5.93%, respectively (χ23 = 26.22, p < 0.001)	24
Edens et al. 33	Compared a national group of homeless and non-homeless VA mental health services users to determine risk and protective factors for homelessness	Cross-sectional, 2009	National sample, N = 118,459; gender analysis completed but not stated; race and ethnicity not stated	Alcohol use disorder Bipolar disorder Drug use disorder GAD MDD Pathological gambling Personality disorder PTSD Schizophrenia	Veterans with severe mental illness had a significantly higher homelessness risk than Veterans without homeless diagnosis Risk of homelessness among Veterans was as follows: • AUD, OR = 1.8, p < 0.0001 • Bipolar disorder, OR = 1.7, p < 0.0001 • Drug use disorder, OR = 4.4, p < 0.001 • GAD, OR = 0.9, p < 0.001 • MDD, OR = 1.4, p < 0.0001 • Pathological gambling, OR = 2.4, p < 0.01 • Personality disorder, OR = 2.2, p < 0.0001 • PTSD, OR = 1.0, not significant • Schizophrenia, OR = 3.3, p < 0.0001 • (CIs were not provided.)	20
Elbogen et al. 32	Examined relationships between financial well-being and community reintegration in Iraq and Afghanistan Veterans	Cross-sectional, 2009	National sample, N = 1,388; male, 67%; Caucasian 70%	MDD PTSD	Veterans who had mental health conditions (e.g., MDD or PTSD) and could not meet basic needs were more likely to experience homelessness than Veterans who had probable mental health conditions and could meet basic needs (χ2 = 11.437, p = 0.0007).	18
Elbogen et al. 34	Investigated empirical association between psychosocial protective factors and subsequent SI among Veterans	Longitudinal, 2009–2011	National sample, N = 1,090; gender not stated; white, 71%	SI	Homelessness in the past year predicted SI among Veterans at wave 2 (χ21 = 12.89, p = 0.0003)	14
Holliday et al. 19	Examined effects of homelessness and justice involvement on mental health outcomes among post-9/11 men and women Veterans	Cross-sectional 2018	National sample, N = 2,227; male, 62%; white, 52.8%	MDD PTSD SI	For men and women Veterans, homelessness associated with more severe mental health symptoms and increased report of SI and attempt compared with those with no history of homelessness.	25

CI = confidence interval; GAD = generalized anxiety disorder; MDD = major depressive disorder; MQQ = Methodological Quality Questionnaire; OR = odds ratio; PTSD = posttraumatic stress disorder; SI = suicidal ideation; SUD = substance use disorder; VA = Veterans Affairs

Table 2.

Results of individual studies for employment and finances

Study	Study purpose	Study design and years	Sample size and characteristics	Mental health condition	Findings	MQQ score
Amick et al. 36	Examined associations between employment and mental health conditions, including PTSD and MDD	Cross-sectional, 2009–2013	National sample, N = 48,821; male, 94%; race and ethnicity not stated	MDD PTSD	PTSD + MDD group had the greatest risk for unemployed-not looking category PTSD + MDD group risk for unemployment, crude RR = 1.44, 95% CI = 1.31–1.58, p < 0.001)	26
Blakey et al. 35	Explored well-being among Veterans with PTSD and SUD and compared them with Veterans with PTSD only, SUD only, or neither disorder	Cross-sectional, 2009–2010	National sample, N = 1,102; male, 82.9%; white, 69.4%	PTSD SUD	Veterans with PTSD and SUD scored lower on vocational and financial well-being compared with Veterans with neither or just one disorder Vocation: χ23 = 36.32, p < 0.001 Finances: χ23 = 92.26, p < 0.001	24
Elbogen et al. 32	Examined relationships between financial well-being and community reintegration in Iraq and Afghanistan Veterans	Cross-sectional, 2009	National sample, N = 1,388; male, 67%; Caucasian, 70%	MDD PTSD	Veterans who had mental health conditions (e.g., MDD or PTSD) were 1) less likely to have money to cover basic needs (e.g., food, clothing, housing, medical care, social activities, and transportation) (χ2 = 87.70, p < 0.0001) and 2) more likely to have lost a job (χ2 = 36.32, p < 0.0001)	18
Elbogen et al. 34	Investigated empirical association between psychosocial protective factors and subsequent SI among Veterans	Longitudinal, 2009–2011	National sample, N = 1,090; white, 71%; gender not stated	SI	Unemployment predicted SI at wave 2 (χ21 = 23.17, p < 0.0001)	14
Possemato et al. 39	Investigated which deployment factors predicted PTSD symptom severity independent of combat trauma in a sample of OEF-OIF Veterans	Cross-sectional, years not stated	Regional sample, N = 150; male, 87%; white, 77%	PTSD	Correlation between employment and PTSD symptom severity (r = −0.19, p < 0.05) Employment independently predicted PTSD symptom severity (β = −0.15, p < 0.05)	22
Sairsingh et al. 38	Examined social factors as they related to MDD among women who served in OEF-OIF	Cross-sectional, 2015–2016	Regional sample, N = 128; female, 100%; white, 82%	MDD	Comfortable financial situation was associated with lower MDD symptoms compared with Veterans who were unable to make ends meet (β = −6.55, p < 0.01) Part-time employment was associated with higher MDD symptoms than full-time employment (β = 3.23, p < 0.01)	19
Sripada et al. 37	Investigated correlations between social networks and support, mental health conditions, and use of mental health services in National Guard Veterans	Cross-sectional, 2011–2013	Regional sample, N = 1,448; male, 91.8%; white, 83%	GAD MDD PTSD SI	No statistically significant association was found between income and probable mental health condition when controlling for other socio-economic conditions	24
Zivin et al. 41	Evaluated relationship between mental health and employment status	Cross-sectional, 2004–2005	National sample, N = 98,867; gender and race and ethnicity were included in analysis but not reported	Bipolar disorder GAD MDD PTSD	Compared with employed patients, VA patients who were unemployed were more likely to have the following mental health conditions:	23
				Schizophrenia SUD	• Schizophrenia (OR = 2.23, CI = 1.50–3.31) • SUD (OR = 2.19, CI = 1.80–2.26) Findings for other mental health conditions were not statistically significant.
Zivin et al. 40	Examined experiences of employment and MDD among Veterans of conventional working age	Longitudinal, 2003–2005	Regional sample, N = 516; male, 94%; white, 83%	MDD	As MDD increased, the odds of employment decreased. Higher mean baseline MDD scores were associated with a decreased likelihood of being employed (OR = 0.94, 95% CI = 0.89–0.99)	25

χ = slope of the regression line; CI = confidence interval; GAD = generalized anxiety disorder; MDD = Major depressive disorder; MQQ = Methodological quality questionnaire; OEF-OIF = Operation Enduring Freedom-Operation Iraqi Freedom; OR = odds ratio; PTSD = posttraumatic stress disorder; RR = relative risk; SI = suicidal ideation; SUD = substance use disorder; VA = Veterans Affairs

Table 3.

Results of individual studies for social support

Study	Study purpose	Study design and years	Sample size and characteristics	Mental health condition	Findings	MQQ score
Avery et al. 51	Evaluated the buffering and direct effect of theories of social support and PTSD	Cross-sectional, years not stated	Regional sample, N = 69; male, 91.3%; Caucasian, 65.2%	PTSD	Post-deployment social support was associated with lower symptoms of PTSD (β = −0.59, p < 0.001)	20
Blakey et al. 35	Explored well-being among Veterans with PTSD and SUD compared with Veterans with PTSD only, SUD only, or neither disorder	Cross-sectional, 2009–2010	National sample, N = 1,102; male, 82.9%; white, 69.4%	PTSD SUD	Veterans with PTSD and SUD scored lower on social wellbeing than Veterans with neither or just one disorder Social support: χ23 = 168.60, p < 0.001	24
Ciarleglio et al. 52	Explored how stress exposures before, during, and after return from deployment influence long-term mental health outcomes	Cross-sectional, 2003–2014	National sample, N = 371; male, 94.9%; Caucasian, 70.4%	GAD MDD PTSD Problem drinking	Post-deployment social support was associated with lower odds of PTSD, MDD, and GAD Social support: • GAD, OR = 0.95, 95% CI = 0.93–0.98, p < 0.001 • MDD, OR = 0.92, 95% CI = 0.89–0.95, p < 0.001 • PTSD, OR = 0.92, 95% CI = 0.89–0.95, p < 0.001 • Problem drinking, OR = 0.98, 95% CI = 0.95–1.0, p = 0.092	25
DeBeer et al. 53	Evaluated whether post-deployment social support moderated the influence of PTSD and MDD symptoms on SI among Veterans	Cross-sectional, time period not stated	Regional sample, N = 130; male, 84.6%; non- Hispanic white, 63.4%	MDD PTSD SI	When social support was low, PTSD and MDD symptoms were positively associated with SI (ΔR2 = 0.07, β = −0.27, p < 0.01)	27
Duax et al. 49	Examined associations among levels of social support and screening positive for PTSD	Cross-sectional, 2006–2009	Regional sample, N = 536; male, 90.3%; white, 79.10%	PTSD	Increase in social support was associated with a reduction in the odds of screening positive for PTSD (OR = 0.92, 95% CI = 0.87–0.96, p < 0.001)	15
Elbogen et al. 34	Investigated empirical association between psychosocial protective factors and subsequent SI among Veterans	Longitudinal, 2009–2010	National sample, N = 1,090; gender breakdown not stated; white, 71%	SI	Social support was a protective factor and predicted lower rates of SI in Veterans (χ 21 = 23.09, p < 0.001)	14
Jakupcak et al. 46	Examined whether PTSD diminishes the buffering effects of social support on elevated suicide risk among Veterans seeking mental health treatment	Cross-sectional, 2004–2007	Regional sample, N = 431; male, 88.9%; white, 66.4%	PTSD SI	Satisfaction with social networks reduced odds of suicide risk. Predictors of elevated SI: • Social networks (OR = 0.18, 95% CI = 0.06–0.50, p < 0.01) • PTSD (OR = 9.37, 95% CI = 2.97–29.51, p < 0.01) • PTSD × Social networks (OR = 3.38, 95% CI = 1.13–10.10, p < 0.05)	21
James et al. 54	Examined the contribution of pre-deployment life events, combat experience, perceptions of threat, and post-deployment social support on mental health symptoms	Longitudinal, time period not stated	Regional sample, N = 271; male, 85%; white, 68%	Alcohol abuse MDD PTSD	Social support provided a highly significant buffer against symptoms of PTSD and MDD • At time point 3: PTSD symptoms (β = −0.42, p < 0.002) • MDD symptoms (β = −0.38, p < 0.002) No significant association was found between alcohol abuse and social support.	22
Lee et al. 50	Examined prediction of long-term PTSD symptom course among a nationwide sample of Veterans	Longitudinal, time period not stated	National sample, N = 1,353; gender and race and ethnicity not stated	PTSD	Social support buffer effects of PTSD symptoms over the course of 20 y Predictor of PTSD symptoms: social support (intercept = −0.56, SE = 0.20, p < 0.05, linear slope = −0.01, SE = 0.02)	22
Lee et al. 55	Examined factors that contribute to PTSD symptomatology among returning Veterans presenting with at least subthreshold PTSD symptoms	Longitudinal, time period not stated	Regional sample, N = 150; male, 88%; white, 83%	PTSD	Social support lowered PTSD symptom severity at Time 1 (β = −0.26, p = 0.004) Social support had no significant change on PTSD symptom severity at time 2 (β = 0.01, p = 0.953)	16
Lemaire & Graham56	Examined factors associated with SI among returning war Veterans	Cross-sectional, 2004–2008	Regional sample, N = 1,740; male, 86.5%; white, 47.1%	SI	Social support lowered the odds of SI (OR = 0.95, 95% CI, 0.92–0.98, p < 0.001)	24
Pietrzak et al. 28	Assessed whether social support attenuates PTSD and MDD symptoms	Cross-sectional, 2003–2007	Regional sample, N = 272; male, 90%; white, 89.4%	MDD PTSD	Increased social support was associated with lesser PTSD and MDD symptoms: • MDD symptoms (β = −0.23, t = 3.17, p < 0.001) • PTSD symptoms (β = −0.31, t = 4.55, p < 0.001)	23
Pietrzak et al. 27	Assessed associations between types of social support and mental health symptom severity			MDD PTSD	Social support was correlated with PTSD (r = −0.56, p < 0.001) and MDD (r = −0.53, p < 0.001)	19
Pietrzak et al. 26	Described correlations between a range of psychometric measures and SI in post-9/11 Veterans			SI	Social support was associated with lower odds of SI (OR = 0.87, 95% CI = 0.76–0.99)	13
Pietrzak & Southwick (2011) 29	Identified types of combat exposure and PTSD symptom clusters and examined how social support differed in these clusters			Alcohol abuse MDD PTSD	Three clusters emerged: controls, PTSD, and resilient; the PTSD cluster had lower social support scores. The resilient group was more likely to score higher on social support than the PTSD group (Cohen’s d = 0.62, p < 0.001)	20
Pietrzak et al. (2011) 57	Examined psychometrics and coping strategy correlates of SI in treatment-seeking OEF-OIF Veterans	Cross-sectional, time period not stated	Regional sample, N = 167; male, 95.8%; white, 63.5%	SI	SI endorsers were more likely to score lower on social support (Cohen’s d = 0.81, p < 0.001)	21
Possemato et al. (2014) 39	Investigated which deployment factors predicted PTSD symptom severity independent of combat trauma in a sample of OEF-OIF Veterans	Cross-sectional, time period not stated	Regional sample, N = 150; male, 87%; white, 77%	PTSD	Social support independently predicted PTSD symptom severity (β = −0.32, p < 0.01)	22
Price et al. (2013) 44	Examined relationships between social support and pretreatment PTSD symptom severity in a group of treatment-seeking Veterans	Randomized controlled trial, time period not stated	Regional sample, N = 69; male, 91%; white, 46%	PTSD	Positive social support was significantly associated with lower PTSD symptom severity (β = 0.45, p < 0.05); the other types of social support (e.g., emotional, tangible, affectionate) were not statistically significant.	21
Proescher et al. (2022) 18	Examined impact of perceived social support on mental health in combat Veterans after military deployment	Cross-sectional, 2012–2017	Regional sample, N = 139; male, 83.4%; white, 34%	GAD MDD PTSD	High perceived social support group reported fewer symptoms, clinical-rated diagnosis of GAD, MDD, and PTSD, or both relative to low perceived social support group: • GAD, χ 22 = 18.87, p < 0.001 • MDD, F2, 135 = 9.26, p < 0.001) • PTSD, FF2, 136 = 4.62, p = 0.011	25
Sairsingh et al. (2018) 38	Examined social factors as they relate to MDD among women who served in OEF-OIF	Cross-sectional, 2015–2016	Regional sample, N = 128; female, 100%; white, 82%	MDD	Social support predicted lower MDD symptoms. Model 1: β = −0.27, p < 0.001 Model 2: β = −0.39, p < 0.001	19
Simons et al. (2020) 58	Estimated a network model of SI among Veterans to advance understanding of the complex array of factors associated with SI	Cross-sectional, time period not stated	Regional sample, N = 276; male, 86%; white, 82%	SI	Social support and SI had a moderate inverse association (r = −0.43, p < 0.001).	12
Sripada et al. (2015) 37	Investigated correlations between social networks and support, mental health conditions, and use of mental health services in National Guard Veterans	Cross-sectional, 2011–2013	Regional sample, N = 1,448; male, 91.8%; white, 83%	GAD MDD PTSD SI	High perceived social support was associated with a lower likelihood of having a probable mental health condition (e.g., GAD, MDD, PTSD, or SI; OR = 0.90, 95% CI = 0.88–0.92, p < 0.001).	24
Zivin et al. (2012) 40	Examined experiences of employment and MDD among Veterans of conventional working age	Longitudinal, 2003–2005	Regional sample, N = 516; male, 94%; White, 83%	MDD	Social support had no statistical relevance in MDD when compared on employment status at baseline, 7 mo, or 18 mo.	25

β = slope of the regression line; CI = confidence interval; GAD = generalized anxiety disorder; MDD = major depressive disorder; MQQ = Methodological quality questionnaire; OEF/OIF = Operation Enduring Freedom/Operation Iraqi Freedom; OR = odds ratio; PTSD = posttraumatic stress disorder; RR = relative risk; SI = suicidal ideation; SUD = substance use disorder; VA = Department of Veterans Affairs.

RESULTS

Twenty-five unique studies examined the association between modifiable social determinants of health and mental health conditions, along with suicidal ideation, among post-9/11 Veterans who served in the U.S. Armed Forces. Across the studies, most Veterans in the aggregate sample were men (mean = 82.7%) and non-Hispanic white (mean = 68.8%). The presence or absence of mental health conditions or concerns was determined for each study using a range of operational methods such as diagnostic codes collected from study participants’ electronic medical records or survey results using validated tools such as the Patient Health Questionnaire-9 (PHQ-9) for depression and Generalized Anxiety Disorder-7 (GAD-7) for anxiety. Several studies investigated more than one modifiable social determinant of health. Thus, to synthesize the nature of the associations, the authors organized study findings by the modifiable social determinant the studies investigated: housing stability, employment or finances, and social support.

Housing stability

Five studies examined the association between housing and mental health among post-9/11 Veterans (Table 1). Four studies used a cross-sectional design, and one study was longitudinal. Most took place between 2009 and 2011 which was shortly after the 2008 housing crisis.³¹ All studies used a national sampling pool, and three of five studies sampled from sources external to the Veterans Health Administration (VHA).

The measures used to examine housing stability consisted primarily of yes-or-no responses to questions such as whether the Veteran had experienced stable housing in the past year. Of the studies that asked about housing stability, one reported information regarding instrument validation.³² Another study operationalized homelessness as use of Department of Veterans Affairs (VA) homeless services or the International Classification of Diseases diagnostic code for homelessness and housing instability as indicated in the electronic medical records.³³

The four cross-sectional studies identified significant associations between housing instability and greater prevalence of mental health conditions, such as PTSD, substance use disorder, and schizophrenia. The studies, including one longitudinal study, also found that Veterans who had experienced housing instability were more likely to report suicidal ideation than Veterans who had stable housing.^19,34

Employment and finances

Nine studies investigated the association between employment and finances and mental health (Table 2). Of these, seven studies used a cross-sectional design, and the remaining two were longitudinal. Five studies included data collected between 2009 and 2013,^32,34–37 which coincides with the Great Recession. More than half of these studies (56%) used national sampling pools. Studies that were conducted regionally had wide geographical representation that included the Northeast,^38,39 Tennessee,³⁹ and the Midwest.³⁷ Of the nine studies, four sourced their samples from within the VHA.^36,39–41

Employment and finances were operationalized differently across the studies. In some studies, questions about financial well-being indirectly included access to food, transportation, and medical care by inquiring about a Veteran’s ability to cover the cost of these basic needs.^32,34,35 Others had employment listed as a covariate in the final models but did not state how the variable was operationalized.^37,39 Three articles categorized employment into groups such as unemployed, homemaker, student, part-time work, or full-time work.^36,40,41 One article dichotomized employment as yes or no.³⁸

Because each study differed slightly in its operational definition of employment and finances, a direct comparison of results was challenging. Overall, studies found that employment and financial stability were associated with fewer mental health symptoms or lower prevalence of mental health conditions. Only one study found a lack of significant association between income and likelihood of having a mental health condition when factors such as social support, hazardous alcohol use, and combat exposure were considered.³⁷ Another interesting result came from an all-women-Veterans study, which found that women who were employed part time were significantly more likely to be depressed than those who were employed full time or unemployed.³⁸ The same study found that women who felt financially comfortable had fewer symptoms of depression than those who were not able to make ends meet. The longitudinal studies reinforced the association between employment and financial stability and mental health and indicate a likely bidirectional relationship. One of the two longitudinal studies found that Veterans who had the financial ability to meet basic needs (e.g., food, transportation) had decreased risk for suicidal ideation.³⁴ The other longitudinal study found that, as depression symptoms improved, Veterans had increased odds for finding and maintaining employment.⁴⁰

Social support

A total of 23 articles representing 20 unique studies investigated the association between social support and mental health (Table 3). Of these, 1 was a randomized controlled trial, and 5 used a longitudinal design. The remaining studies were cross-sectional. Five of the 20 studies sourced participants external to the VA, with 3 of these studies using a national sample. The remaining studies sampled from local VA Medical Centers, such as those based in Central Texas, Chicago, and Puget Sound.

The majority of studies (n = 13) operationalized social support using the post-deployment social support module from the validated Deployment Risk and Resilience Inventory.^42,43 Two other studies used the Medical Outcomes Study Social Support Survey,^40,44 which is also a widely used, validated scale.⁴⁵ Other studies assessed social support using the Quality of Life Interview or the Multidimensional Scale of Perceived Social Support.^18,46 Both instruments are validated.^47,48 The remaining studies inquired about social support using questions that pertained to endorsing at least one close and secure relationship with someone who helps during times of stress and the emotional support they get from people other than their family,³⁵ scaled questions about sources of support,⁴⁹ and assessment of satisfaction with emotional support received from family and friends.³⁴ Information regarding the validity of these questions was not reported in the study articles.

The association between social support and PTSD was examined in 18 of the 20 studies. These studies consistently reported an inverse relationship between social support and symptoms of PTSD, with 3 longitudinal studies reporting that social support protects Veterans from symptoms of PTSD by moderating symptoms over time.⁵⁰ With respect to suicidal ideation, 4 studies reported an inverse association between social support and thoughts of suicide, with a longitudinal study finding that greater social support predicted lower rates of suicidal ideation among Veterans.⁵⁰

Risk of bias in studies, reporting biases, and certainty of evidence

The MQQ assessment scores ranged from 12 (Simons et al.)⁵⁸ to 27 (DeBeer et al.),⁵³ with an average score of 21. The most common methodological problems were reliability and validity of data collected, data analysis plan not consistent with the research question or design, and lack of meaningful interpretation of data to link to scientific evidence or theoretical or conceptual framework. Most studies successfully defined the construct or phenomenon of interest theoretically or conceptually, included a statement of how the variables corresponding to the construct were measured, and linked the research design to the research question or hypothesis.

A potential source of bias stems from the number of studies (n = 19) that sourced Veteran samples from the VHA. Veterans who use VHA services tend to have lower incomes and less education than Veterans who do not use VHA services.^59,60 In addition, bias may exist in studies that did not use validated methods of assessing the modifiable social determinants. Finally, although all studies reported statistical findings, some used statistics such as correlations that did not allow for control of covariates. More robust statistical analysis would have provided greater clarity on the associations of interest.

DISCUSSION

The aim of this systematic review was to assess the current state of knowledge of the associations between modifiable social determinants and mental health among U.S. Veterans who served post-9/11. Although 25 unique studies met the inclusion criteria, they examined a narrow range of modifiable social determinants of health: housing stability, employment and financial stability, and social support. Findings from these studies were consistent overall: Veterans with compromised social determinants had poorer mental health. The handful of longitudinal studies identified in the systematic review suggest that these modifiable social determinants have a protective effect on mental health.

The majority of studies identified through the systematic search focused on the association between social support and PTSD. Most of these studies used validated measures for these two constructs and thus provide robust evidence to support incorporating social support into PTSD treatment and maintenance plans. Indeed, the VHA has peer specialists who help Veterans with mental health disorders while also providing social support.^61–63 Community-based programs such as Vet Centers are another example of integration of social support and mental health care.⁶⁴ The evidence in this systematic review supports continued development of these types of networking programs, which may be particularly helpful for Veterans who may not be ready to engage in formal mental health treatment.¹⁸

Study quality

Overall, the studies were of moderate to high quality, based on the mean and range of MQQ scores. Specifically, the methodological quality of the studies of social support was rated as good to excellent. The quantity and quality of the evidence identified for social support’s association with PTSD did not extend to the other modifiable social determinants identified in the systematic review. Although some of the questions that asked about housing stability and employment and financial stability appeared to have been validated, there was an overall lack of psychometric validation for questions that assessed these modifiable social determinants. This challenge is not unique to the studies in this review; in general, measures for social determinants lack robust psychometrics.⁶⁵

Of note, the MQQ scale gives each criterion equal weight, even though some criteria may compromise the study quality more than others because of the degree of bias introduced. For example, studies that lacked validated assessments of the modifiable social determinant may not have accurately measured the intended concept. Other studies presented ambiguous data analysis; thus, it was difficult to determine the quality of the findings presented.

Implications and future directions

Evidence found in this systematic review suggests that modifiable social determinants are associated with mental health. The review also identified significant gaps in research that need to be addressed. First, more studies are needed that examine a broader range of modifiable social determinants of health such as food security, education, housing quality, neighborhood safety, child care, and transportation.²² The scarcity of studies examining a broader range of social determinants may reflect a lack of awareness of the importance of these factors to health and well-being. Educating health care providers and those who conduct research with Veterans will help to close this gap.⁶⁶

Studies also need to be more inclusive of women Veterans and those with a minority background. Most of the sampled Veterans were non-Hispanic white men. Of note, the two studies found during the updated search had more diverse samples, with one study oversampling for women and the other study purposively sampling minorities.^18,19 Racial and gender disparities in access to health care and health outcomes are well established,⁶⁷ and adverse social determinants are key drivers of these disparities.^68,69 Future studies should prioritize recruitment of women and racial and ethnic minorities, using methods such as purposive sampling,⁷⁰ to begin to address this crucial issue.

More important, future studies need to use more comprehensive methods and validated instruments to produce more robust evidence. Although most studies that examined social support in relation to mental health used validated instruments, questions used to assess housing stability and employment and finances lacked similar rigor. Some social determinants have robust measures ready for use. For example, food security is often measured using the robustly validated U.S. Household Food Security Survey Module.⁷¹ Other social determinants, such as employment and finances, child care needs, and transportation, need validated questions.⁷² Having validated questions to measure these social determinants may help researchers incorporate these factors into their studies. Consistent use of these questions would increase the quality and quantity of evidence on the association between modifiable social determinants and mental health among Veterans.

Finally, future studies should examine associations between modifiable social determinants and mental health using a longitudinal design. The small number of longitudinal studies included in this systematic review support the notion that positive social determinants are protective of mental health. However, additional longitudinal research, with more robust measures and methods, will provide the evidence needed to support changes to clinical care and policy reform.

Limitations

This systematic review has some limitations. First, it focused on post-9/11 Veterans who served in the U.S. Armed Forces and excluded other relevant literature focused on Veterans from prior service eras and nations. However, findings are similar to what has been found among other Veteran populations.^5,73,74 Second, some studies that had evidence on the association between employment and finances and mental health actually used these variables as covariates. These studies were identified in the systematic search because they were studying social support. Thus, it is possible that other studies in which these, or other modifiable social determinants such as homelessness, were used as covariates may not have surfaced in the search. Third, this systematic review was limited to publications in peer-reviewed journals. This may have introduced publication bias in which studies with null findings may not have been identified.

Conclusion

Evidence suggests the need to consider modifiable social determinants of health when designing mental health interventions. However, more research that encompasses a wider range of modifiable social determinants such as food security, education, and transportation using comprehensive methods and validated instruments is needed. Future research also needs to intentionally include Veterans from diverse gender and racial and ethnic groups.

Biographies

Nipa Kamdar, PhD, RN, FNP-BC, is a research science specialist at the Center for Innovations in Quality, Effectiveness, and Safety, a joint research lab between Michael E. DeBakey Veterans Affairs Medical Center and Baylor College of Medicine. Her research on food insecurity and related social needs builds on her experiences as a family nurse practitioner serving communities with low resources.

Sundas Khan, MD, is a quantitative methodologist at the Center for Innovations in Quality, Effectiveness, and Safety, a joint research lab between Michael E. DeBakey Veterans Affairs Medical Center and Baylor College of Medicine. Her research expertise lies in implementation of evidence-based medicine, adaptive clinical decision support, and developing electronic tools for patients and providers based on user-centered design. In her current role, Khan is responsible for informing research study design, including quantitative and qualitative research methodologies.

Diana P. Brostow, PhD, MPH, RDN, is a clinical research nutritionist at the Rocky Mountain Mental Illness Research Education and Clinical Center at the Rocky Mountain Regional Veterans Affairs Medical Center and Assistant Professor at the Department of Physical Medicine and Rehabilitation, University of Colorado School of Medicine. Her research focuses on psychosocial determinants of food insecurity and dietary behaviors.

Lia Spencer is an undergraduate student at Brandeis University, studying biology and public health. She is very interested in the intersections among social determinants, mental health, and physical health. She plans to apply to medical school after her anticipated graduation in May 2023.

Sharmily Roy, MPH, is a health services-health economics PhD student at the School of Public Health, University of Texas Health Science Center at Houston. She received her MPH at Georgia State University and her BA in international relations from Agnes Scott College in Atlanta, Georgia. Her background includes public health program monitoring and evaluation, and she is interested in research to improve health equity and access.

Amy Sisson, MS, MLS, is a medical librarian at The Texas Medical Center (TMC) Library. In addition to her library master’s degree from the University at Albany, State University of New York, she holds an MS degree in space studies from the University of North Dakota. She serves as a TMC Library liaison to the Baylor College of Medicine, including its partner program with the Michael E. DeBakey VA Medical Center, and specializes in systematic review literature searching.

Natalie E. Hundt, PhD, is a clinical psychologist at the Michael E. DeBakey Veterans Affairs Medical Center and Associate Professor in the Department of Psychiatry and Behavioral Sciences at Baylor College of Medicine in Houston, Texas. Her research interests focus on access to mental health treatment and implementation of evidence-based psychotherapies.

APPENDIX

Appendix

APPENDIX. MEDLINE Ovid Search Strategy

1. Veterans/

2. (veteran* or “Wounded Warrior*”).ti,ab,kw.

3. ((former* or prior* or past) adj3 (military or “service member*”)).ti,ab,kw.

4. (“former* military” or “former* service member*” or “prior* military” or “prior* service member*” or “past military” or “past service member*”).kw.

5. or/1–4

6. September 11 Terrorist Attacks/

7. (“September 11*” or “Sept. 11*” or “Sept 11*” or “9/11”).ti,ab,kw.

8. (“operation enduring freedom” or OEF or “operation iraqi freedom” or OIF or “operation new dawn” or OND or “operation inherent resolve” or OIR).ti,ab,kw.

9. (war* adj2 (afghanistan* or iraq*)).ti,ab,kw.

10. (“afghanistan war” or “iraq* war*”).kw.

11. or/6–10

12. “Social determinants of health”/

13. (social* adj3 (need* or determinant* or stab* or instab* or unstab* or capital*)).ti,ab,kw.

14. (“social* need*” or “social* determinant*” or “social* stab*” or “social* instab*” or “social* unstab*” or “social* capital*”).kw.

15. Housing/ or Public Housing/ or Homeless Persons/

16. (housing* or housed* or unhoused* or homeless* or “home-less*”).ti,ab,kw.

17. Electricity/ or Water/ or Heating/ or Natural Gas/

18. (electricity or water or heat* or “natural gas” or utility or utilities).ti,ab,kw.

19. exp Environmental Pollution/

20. (pollution* or “air qualit*”).ti,ab,kw.

21. Food Insecurity/ or Food Security/ or Hunger/ or Nutritional Status/ or Malnutrition/

22. (food* or hunger* or nutrition* or malnutrition* or malnourish*).ti,ab,kw.

23. Social Support/ or Psychosocial Support Systems/

24. ((social* or psychosocial* or family* or families* or communit* or peer*) adj3 support*).ti,ab,kw.

25. (“social support*” or “psychosocial support*” or “family support*” or “community support*” or “peer* support*”).kw.

26. Employment/ or Unemployment/

27. (employ* or unemploy* or underemploy* or “job assist*” or “job security*”).ti,ab,kw.

28. Poverty/ or Financial Stress/

29. (poverty* or impoverish*).ti,ab,kw.

30. (financial* adj3 (stab* or instab* or unstab* or strain* or stress* or securit* or insecurit* or status*)).ti,ab,kw.

31. (“financial* stab*” or “financial* instab*” or “financial* unstab*” or “financial* strain*” or “financial* stress*” or “financial securit*” or “financial insecurit*” or “financial status*”).kw.

32. (foreclos* or “fore-clos*” or forclos* or “for-clos*”). ti,ab,kw.

33. exp Socioeconomic Factors/

34. ((socioeconom* or econom* or social*) adj3 (factor* or class* or status* or level* or advantage* or disadvantage*)).ti,ab,kw.

35. (“socioeconom* factor*” or “socioeconom class*” or “socioeconom* status*” or “socioeconom* level*” or “socioeconom* advantage*” or “socioeconom* disadvantage*” or “econom* factor*” or “econom* class” or “econom* status*” or “econom* level*” or “econom* advantage*” or “econom* disadvantage*” or “social factor*” or “social class*” or “social status*” or “social level*” or “social advantage*” or “social disadvantage*”).kw.

36. (income* adj3 (equal* or inequal* or inequit* or level* or status*)).ti,ab,kw.

37. (“income equal*” or “income inequal*” or “income inequit*” or “income level*” or “income status*”).kw.

38. exp Education/

39. educat*.ti,ab,kw.

40. Internal-External Control/

41. (sense* adj3 (mastery* or control*)).ti,ab,kw.

42. (“sense of mastery” or “sense of control”).kw.

43. exp Violence/

44. violen*.ti,ab,kw.

45. (“military sexual trauma” or MST).ti,ab,kw.

46. exp Safety/

47. safety*.ti,ab,kw.

48. exp Transportation/

49. transportation*.ti,ab,kw.

50. (“spatial* mismatch*” or “spacial mis-match*”). ti,ab,kw.

51. or/12–50

52. Mental Health/

53. (mental* adj3 (health* or “well-being*” or wellbeing* or well* or unwell*)).ti,ab,kw.

54. (“mental* health*” or “mental* well-being*” or “mental* wellbeing*” or “mental* well*” or “mental* unwell*”). kw.

55. exp Psychotic Disorders/

56. (psych* adj2 (disorder* or “disorder*”)).ti,ab,kw.

57. (“psych* disorder*” or “psych* dis-order*”).kw.

58. (psychosis* or psychoses*).ti,ab,kw.

59. exp Stress, Psychological/

60. (stress* or burnout* or “burn-out*” or “burned out*”). ti,ab,kw.

61. Stress Disorders, Post-Traumatic/

62. (“post-trauma*” or posttrauma* or PTSD).ti,ab,kw.

63. Depression/ or exp Anxiety/ or exp Anxiety Disorders/

64. (depress* or anxiet* or anxious*).ti,ab,kw.

65. exp Substance-Related Disorders/

66. ((substance* or drug* or alcohol*) adj2 (abus* or misuse* or “mis-use*” or disorder* or “dis-order*” or dependenc*)).ti,ab,kw.

67. (“substance* abus*” or “substance* misus*” or “substance* mis-us*” or “substance* disorder*” or “substance dis-order*” or “substance dependenc*” or “drug* abus*” or “drug* misus*” or “drug* mis-us*” or “drug* disorder*” or “drug dis-order*” or “drug dependenc*” or “alcohol* abus*” or “alcohol* misus*” or “alcohol* mis-us*” or “alcohol* disorder*” or “alcohol* dis-order*” or “alcohol dependenc*”).kw.

68. (alcoholic* or alcoholism* or “binge* drink*” or addict*).ti,ab,kw.

69. exp Suicide/

70. suicid*.ti,ab,kw.

71. or/52–66

72. 5 and 11 and 51 and 71