Firearm Suicide Among Veterans of the U.S. Military: A Systematic Review

ABSTRACT

Introduction

In the United States, firearm suicide represents a major cause of preventable, premature death among veterans. The purpose of this systematic review was to characterize the body of literature on veteran firearm suicide and identify areas for future research, which may facilitate the development of firearm suicide interventions in Veterans Health Administration (VHA) and non–Veterans Health Administration clinical settings.

Materials and Methods

All randomized controlled trials, quasi-experimental, naturalistic, observational, and case study designs published between January 1, 1990 and February 21, 2019 were included in our review. Following title and abstract review, 65 papers were included in our full-text review and 37 studies were included in our analysis. We based our approach on a modification of the Preferred Reporting Items for Systematic Reviews and Meta-Analysis guidelines. Studies were grouped into broad, nonmutually exclusive categories: (1) heterogeneity of datasets and veteran status determination for inclusion, (2) service histories, (3) firearm ownership, storage, behaviors, and risk perceptions, (4) patient and clinician attitudes toward firearm restriction interventions, (5) firearm suicide risk factors by study population, and (6) assessments of clinical firearm interventions.

Results

This body of literature consists predominately of cross-sectional studies with mixed definitions and validation of veteran status, which revealed high concordance of increased risk of firearm suicide compared with nonveterans. Veterans have higher rates of firearm ownership than the general population, primarily citing personal protection as the reason for gun ownership. Veterans often exhibit risky firearm usage and storage behaviors but tend to favor measures that limit access to firearms by at-risk individuals. Despite this, there remains persistent hesitation among clinicians to screen and counsel veterans on firearm safety.

Conclusions

This systematic review highlights an urgent need to produce higher quality evidence and new data with standard definitions that are critical to inform clinical practice and enhance public health measures to reduce firearm suicide among veterans.

INTRODUCTION

In the United States, veterans are disproportionately affected by firearm suicide. Veterans account for 5.6%^1,2 of the total U.S. population but 15.6%³ of suicides each year, 69% of which occur by firearms. An estimated 4,260 veterans died by firearm suicide in the United States in 2017, nearly equal to the total number of U.S. all-cause deaths in Iraq since 2003.³ The current age- and sex-adjusted veteran suicide death rate of 27.7 per 100,000⁴ is 1.5 times higher than the general population.⁵ These suicides represent preventable premature deaths in the veteran population.

Suicide may be conceptually divided into a multi-step pathway beginning with suicidal ideation and progressing along a spectrum of intent, planning, and the act of suicide itself.^6,7 The time from an initial thought of attempting suicide to the action of suicide often occurs in a matter of minutes.⁸ Each factor involved in the transition from a nonsuicidal state to suicidal ideation, planning, and a subsequent attempt has important implications. This concept highlights modifiable risk factors present in the progression through these stages and informs clinical and policy approaches to suicide reduction.

Access to lethal means, including firearms, is one such modifiable risk factor. Despite limitations in individual-level suicide prediction, population and individual-level data demonstrate strong and consistent associations between firearm access and subsequent suicide for individuals in high-acuity states.^9–11 Among people in the planning stage of the suicide pathway, firearm owners are more likely to have a plan involving a firearm than nonfirearm owners.¹² The case-fatality rate for firearm suicide is 90%.¹³ Firearms are the most common suicide method in both veterans and nonveterans in the United States,¹⁴ making it an important area of focus for suicide prevention efforts. Veteran suicide prevention is currently a top clinical priority at the DVA,¹⁵ and the DVA has called for prevention efforts to include firearm access in several reports over the past decade.

Despite their importance, firearms remain an understudied component of suicide among veterans.¹⁶ Veterans are exposed to multiple life events during their service that are associated with unique attitudes toward firearms and increased suicide risk. Understanding veterans’ military experiences, service histories, firearm ownership, and attitudes about firearms are essential elements to best inform how to intervene within a culturally distinct group.¹⁷

In this systematic review, we characterized the current literature related to veteran firearm suicide. In so doing, we also identified areas for future research, which may facilitate the development of firearm suicide interventions in Veterans Health Administration (VHA) and non-VHA clinical settings. Furthermore, this review sought to characterize datasets and veteran-specific population characteristics for readers who may be unfamiliar with the military and veteran community.

METHODS

Data Sources and Search Strategy

Using a broad-based query, we systematically searched PubMed, Embase, and PsycINFO for the published, peer-reviewed literature that discusses veterans, firearms, and suicide. The queries used database-specific indexing terms, and synonyms were limited to text searches of titles and abstracts (see Appendix  A).

Study Selection

We sought to include all randomized controlled trials and quasi-experimental, naturalistic, observational, and case study designs published between January 1, 1990 and February 21, 2019. The initial queries produced 685 articles, which were uploaded to DistillerSR (Evidence Partners, Ottawa, Canada).¹⁸ After removing duplicates, 279 were included in the initial review. Titles and abstracts were screened by two reviewers (K.H. and J.T.) with an oversight of a research methodologist (M.B.). The senior reviewer (C.C.) made the final determination on inclusion when there was conflict between first-line reviewers. The title review excluded 45 articles. The abstract review excluded a further 141 articles that were identified as editorials, commentaries, book chapters, review articles, or did not discuss study subjects associated with a U.S. military or veteran population. A total of 93 papers were identified and categorized based on study populations as described in the title and abstract: (1) 40 discussed veterans, (2) 25 made equivocal mention of veterans, and (3) 28 described active duty uniformed services, including the National Guard and reserve component. Excluding the third category, 65 papers were carried forward for the full-text review (Fig. 1).

FIGURE 1.

Preferred Reporting Items for Systematic Reviews and Meta-Analysis flowchart.

Data Extraction and Study Quality Assessment

We developed our approach based on a modification of the Preferred Reporting Items for Systematic Reviews and Meta-Analysis guidelines,^19,20 which are typically used for healthcare intervention and effectiveness studies.

Two reviewers (J.T. and K.H.) independently reviewed the text of each paper to capture the following data elements: (1) authors; (2) year of publication; (3) study question; (4) study design; (5) study population and setting, which was characterized as either veteran or nonveteran participants, inside or outside the VHA system, geographic region of the United States, era of military service, and presence or absence of any military service history items; (6) time period, which corresponds to era of service; (7) data source, which was further characterized as primary data with measurement method (e.g., interview, survey), measurement instrument (e.g., validated screening tools), and sampling methods (e.g., either probability or nonprobability, by subtype); (8) nonresponder analysis for studies with response rates below 70%; (9) outcome measures, as reported by authors, relevant to the study objective and selected population; (10) author characterizations of study findings in terms of informing clinical practices, future research of gaps, or policy implications; (11) study limitations and generalizability of findings; and (12) the funding source or sponsoring institution to assess VHA involvement. We assessed studies for confounding, selection and information bias, and how authors accounted for potential threats to validity. A complete list of studies reviewed and their primary research questions can be found in Appendix  B.

Data Synthesis and Analysis

Studies were grouped into broad, nonmutually exclusive categories: (1) heterogeneity of datasets and veteran status determination for inclusion, (2) service histories, (3) firearm ownership, storage, behaviors, and risk perceptions, (4) patient and clinician attitudes toward firearm restriction interventions, (5) firearm suicide risk factors by study population, and (6) assessments of clinical firearm interventions.

RESULTS

This analysis included 37 papers: 28 cross-sectional studies, 3 longitudinal studies, 3 case–control studies, 2 case studies, and 1 qualitative study. A full list is shown in Appendix 2. Samples were primarily drawn from VHA mental health populations with limited or undefined regional sampling. Across all studies, most data were self-reported during structured interviews or questionnaires and large national or state datasets. Most studies focused on sample descriptive statistics, comparative prevalence of putative risk factors or associations among study subjects, perceptions and attitudes regarding firearms and specific clinical interventions to reduce firearm suicide risk, and behaviors associated with firearm ownership. Service history reporting varied throughout studies and was not easily categorized. The conclusions of most papers included statements calling for improved clinical screening for suicide risk, firearm access, and clinician education.

Heterogeneity of Data Sources, Veteran Status, Prevalence, and Baseline Risk

Primary outcomes of several studies included firearm suicide prevalence, suicidal ideation, and firearm ownership stratifying by or adjusting for veteran status, or sampling from validated veteran populations. Across all studies, there was no unifying definition of “veteran status.” Determination varied by data source and study methodology and was performed in one of three ways: (1) sampling from VHA-enrolled veteran populations and then cross-referencing suicide events from state or national mortality records; (2) querying national or state mortality records for fatal suicides and then cross-referencing DVA or DoD records, death certificates, or next-of-kin reports; or (3) self-reports and administrative records for studies with living participants. For example, Miller et al. analyzed the Cancer Prevention Study-II, using living participant responses to “[having] ever served in the United States Armed Services and service branch.”²¹ Similarly, studies referencing the 2015 National Firearm Survey, Simonetti et al.^22,23 and Cleveland et al.²⁴ used participants’ self-reports. Several studies used death certificates. Kaplan et al. noted 11.4% of suicide decedents had incomplete or unknown military service documented on their death certificate.²⁵ Freeman et al. used self-reports and the DD-214 (official discharge paperwork), to confirm veteran status.^26,27

Overall elevated veteran suicide prevalence and risk, both all cause and firearm specific, is well established by the DVA’s suicide surveillance system.²⁸ Studies comparing firearm suicides among veterans to nonveterans show mixed results (Supplementary Table S1). In summary, elevated risk of firearm suicide among veterans compared with nonveterans ranged from no significant difference in a nonvalidated veteran sample of men aged 45 years and older from 1982 to 2004,²¹ to 6% higher in men and 18% in women in a validated veteran sample from 1999 to 2009,²⁹ to 30% higher in men and 47% women in a nonvalidated veteran sample between 2003 and 2006.³⁰ McCarten et al. note a 16.2% and 75.4% age-adjusted increase in firearm suicides among men and women, respectively, between 2001 and 2010 in a validated veteran sample.³¹ None of these studies are contemporaneously generalizable to the national veteran population because of limited availability of curated state data, which are diverse but geographically limited and have the potential for misclassified veteran status.

Associations Between Suicide and Mental Health

Several associations were explored primarily in small pilot studies and large-scale cross-sectional studies from general veteran populations or specific veteran mental health populations. Firearm suicide and suicidal ideation were treated either as a primary outcome or covariate in studies of all-cause suicide.

Kaplan et al. found that veterans aged 65 years or older were more likely to use a firearm than those aged 18 to 34 years.³² Compared with the same group using psychological autopsies, all-cause suicide decedents aged 35 years and older had strong associations with preexisting health conditions, with adjusted odds ratios (ORs) ranging from 2.41 to 36.09. Nicotine exposure elevated suicide risk of independent of veteran status and firearm ownership in a dose–response fashion.³³

Seyfried et al. found in a sample of 294,952 veterans in the VHA system followed from 2001 to 2005 that firearms accounted for 72.6% of suicides among dementia patients and that 75% of all-cause suicides occurred in those with newly diagnosed dementia.³⁴ The authors note associations of suicide with depression (OR 2.04, 1.45-2.85, P < .01), in-patient psychiatric admission (OR 2.31, 1.54-3.46, P < .001), anti-anxiety medications (OR 1.98, 1.48-2.65, P < .001), and antidepressants (OR 2.11, 1.57-2.85, P < .001).³⁴ In contrast, a brief report by Trivedi et al. found that of 13,881 fatal firearm suicides, 52% had no diagnosed mental illness on record.³⁵

Associations Between Suicide and Other Factors

Regarding relationship stressors, there was increased risk of firearm suicide in samples aged 34 and younger,³⁶ among women (38.6% using a firearm versus nonfirearm 30.4%, P < .001),³⁰ and increased risk of suicidal ideation among those with marital difficulties (OR 5.52, 1.57-19.45, P = .01), most of whom owned firearms.³⁷ A recent intimate partner problem was documented in VHA clinical encounters among 70.7% of veterans with a positive suicide risk assessment.³⁸ Other studies found no association between relationship stressors and firearm suicide in men when compared to nonfirearm suicide³⁰ and a statistically significant decreased risk of a relationship problems among suicide decedents with a cancer diagnosis compared with those without (intimate partner or relationship problems, adjusted odds ratio [AOR] 0.32-0.81)³⁹; notably the latter study did not stratify by veteran status or firearm-specific suicide.

There were increased rates of all-cause and firearm suicide in rural versus urban settings in validated and nonvalidated veteran samples^17,40 and in states with higher firearm ownership and less restrictive firearm laws.⁴¹ States with larger veteran populations tended to have higher numbers of all-cause and firearm suicide, and veterans tended to reside in states with fewer firearm restrictions.⁴² Miller et al. found no difference in suicide when comparing service era cohorts and branches of service in a nonvalidated veteran sample.²¹

Higher intensity combat experiences among a small sample of veterans of Operation Iraqi Freedom (OIF) or Operation Enduring Freedom (OEF) with alcohol use disorder was modestly correlated with increased suicidal ideation (P < .05); 0.29 for firing a weapon at an adversary and 0.25 for killing or believing one killed another human.⁴³ Additional associations with service history are described below.

Firearm Ownership, Attitudes, Behaviors

Firearm ownership prevalence varied across studies (Supplementary Table S2). In five surveys composed of veterans in the general population^22–24 and general VHA mental health and primary care settings,^44,45 firearm ownership was between 44.0% and 44.9%. Smaller studies at single-facility or single-region inpatient settings varied widely. Prevalence of gun ownership among Vietnam, Gulf War, and OIF/OEF veterans at a single PTSD treatment facility was 28%.⁴⁶ In a separate study at two PTSD treatment sites (one VHA and one non-VHA in California), ownership was around 33%, with higher combat intensity associated with increased ownership among OIF/OEF veterans but not among Vietnam veterans.⁴⁷ Another sample of mostly Vietnam combat veterans from an Arkansas-based VHA facility showed that the highest average ownership was among those with PTSD when compared to substance use disorders or schizophrenia,²⁷ with 75% ownership among a separate sample at the same site.²⁶ A separate VHA study of a sample of veterans with opiate use disorder reported an AOR of 3.67 (1.14-11.83, P < .03) comparing ideators with nonideators.³⁷ McGlade et al. reported firearm access, rather than ownership, among 59% of veteran suicide decedents in Utah.⁴⁸ Firearm access compared with civilians was statistically significantly higher among veterans, with an AOR 2.05 (1.03-4.06, P < .04).⁴⁸ Personal protection was the most reported reason for firearm ownership, concordant across all studies that assessed this characteristic, between 58% and 70%.^{22–24,44,49}

Nearly a third of veterans stored a firearm loaded and unlocked as reported in larger surveys.^22,23,44 Veterans with PTSD, compared with those without, in smaller VHA mental health settings had higher likelihood of unsafe storage.⁴⁷ In one study, 25% of veterans with suicide risk factors would remove their firearm from the home if they believed another member of the household was at risk, over a third of respondents felt their firearms were not accessible to anyone else, and only 6.3% were familiar with firearm access and increased suicide risk.²³ Interestingly, the perception of one’s own risk of gun injury among patients with a mental diagnosis tended to be higher at 74.8%.⁵⁰ Four studies assessed veterans’ perspectives of firearm counseling and restriction interventions.^22,44,45,49 A small focus group by Walters et al. assessed attitudes toward firearm counseling among clinicians, VHA administrative leaders, and families of VHA utilizers with high suicide risk.⁴⁹ Clinicians generally agreed that firearm safety counseling is important but hesitated to engage about the topic with patients because of unfamiliarity with guns, rapport concerns, or not wanting to expose family members to potential safety risks involved with firearm removal from a home. Clinicians also cited difficulty in risk-stratifying patients to determine when to return a confiscated or removed firearm.⁴⁹

Valenstein et al. found that 33% of surveyed VHA users, consisting mostly of older Caucasian males, felt any firearm restriction intervention would work,^44,45 and most agreed that educating about general suicide risk to veterans’ family members was appropriate.⁴⁵ Over 90% of the VHA mental health patients surveyed were amenable to one or more firearm restriction interventions, with differences in intervention intensity, and nearly three-quarter was amenable to more restrictive measures.⁴⁵

A subset of studies sought to further characterize patient and clinician interactions, specific relevant portions of which are shown in Supplementary Table S3. Dobscha et al. studied clinicians’ care decisions in response to positive suicide risk assessments by analyzing data from 1,017 OIF/OEF veteran VHA patients’ primary care medical records between 2008 and 2009.³⁸ Of those with a positive suicide risk assessment, only 15% were screened for firearm access based on clinician medical record entries, and only three cases documented any form of firearm restriction counseling. Similarly, Valenstein et al., report only 50% of VHA mental health users with a suicide plan were screened for firearm access, and only 27% of all VHA users in a mental health setting were ever asked about firearms.⁴⁴

DISCUSSION

This review describes the breadth of research on veteran firearm suicide in the last nearly 30 years, but, in so doing, it exposes wide gaps in knowledge and practice. These findings illuminate an urgent need for multidisciplinary, multisector research. First, veterans are at substantially increased risk of firearm suicide. Veterans have higher rates of firearm ownership than the general population, primarily for personal protection. Although many store firearms unsafely, they generally favor measures to restrict access to firearms among at-risk individuals, particularly among those who experienced higher intensity combat and those who reside in regions outside of New England. Data on firearm screening are dated, but there is persistent hesitation among providers to screen and counsel because of a lack of familiarity with guns, little confidence in intervention effectiveness, or philosophical reasons related to constitutional rights. Furthermore, the lack of data describing the proximity of mental health providers and crisis resources to at-risk individuals is a striking limitation in the ability to appropriately risk stratify veterans being evaluated for suicide. This lack of assessment of community resources in rural and urban settings may confound the assessment of statistically significant risk factors associated with population density, which is important to the allocation of resources intended to reduce suicide.

A more standardized approach to this research is necessary to better understand veterans’ specific firearm-related practices and to inform the development of targeted interventions. First, we must use the standard definition of veteran⁴ status and means of validating veteran status. Nearly two-third of veterans of the Global War on Terrorism received their care outside of the VHA system in 2015, 12% and 88% for men and women, respectively.⁵¹ Leveraging third-party digital identity platforms⁵² could improve researchers’ ability to capture this population. Validating sample populations against DVA and DoD records will allow for greater generalizability, but it also adds costs to the studies, limits the number of participants if they do not have access to military service records, or creates an added barrier in surveys with historically high nonresponse rates.

Second, we must integrate service history data into analyses to allow for greater depth of risk and protective factor assessment among veterans. Among the included studies, service history data collected from subjective self-reports on questionnaires and clinical screening tools or objectively sourced from administrative or health records were rarely assessed in studies where the primary outcome was fatal all-cause or firearm suicide. The spectrum of service histories is wide-ranging across all studies but narrow within individual studies, which limits strength in concordance. Further, these data were often included in descriptive statistics of the sample but not assessed as part of the primary study objective to make comparisons. Little is known about service histories, such as discharge status, occupational specialty, or punitive actions among others, nor how they relate to firearm suicide, much of which can be found succinctly on the DD-214 discharge form. A service member’s discharge status may become especially relevant when one considers the attendant risk of mental health problems, substance use disorders, and legal problems, among others that may be seen with greater prevalence among those with other-than-honorable discharges who do not qualify for care within the VHA system. Many authors note this limitation in the body of knowledge. Efforts to combine validated veteran data with DVA or DoD administrative data may yield insights to assist with risk stratifying veterans during screening in clinical settings. It is also important to expand service era assessment to distinguish era versus combat deployment. This can be done with the Deployment Risk and Resilience Inventory, Combat Experiences Scale,⁵³ or systems used to populate the DD-214, depending on the sample being studied. The Deployment Risk and Resilience Inventory is robust and can be piecemealed for focused subsets of questions characterizing wartime exposures.⁵⁴

Service history and other relevant aspects of life experience and psychiatric diagnoses are most comprehensively collected through psychological autopsy.^55,56 Although more expensive, studies utilizing the multiple collateral interviews and extensive record review undertaken for psychological autopsy would dramatically improve our understanding of veteran firearm suicide, even when small samples are revisited for interview. This strategy was successfully pursued most recently as a part of the Army STARRS project, where psychological autopsy on a subset of a large sample revealed that home access to a firearm was strongly associated with suicide among U.S. Army soldiers.⁵⁷ Third, we must standardize geographic boundaries in the context of veteran-specific health services research. It was difficult to report on regional similarities and differences across the studies in this review because of a lack of standard geographic boundaries. There are clear geographic differences in attitudes toward firearm access and use among veterans, providers, and the general population. The importance of crafting clinical, policy, and legislative approaches will require a clear sense of local and regional needs and attitudes. For example, there is strong consensus across studies that personal protection is a primary reason for firearm ownership, and providers must understand the implications of ownership in that context, especially when assessing for suicide. Veteran Integrated Service Networks or ZIP codes are potential approaches to accounting for geography that could support a more robust regional assessment in this work.

There are also limited cross-sectional data characterizing the events preceding fatal and nonfatal firearm suicide attempts. The current included studies suggest that relationship stressors are greatest among those younger than 34 and among women of all age ranges, but is less apparent in older populations. This limitation in the existing literature further emphasizes the importance of psychological autopsies in future research.

From this review, there is a clear need to emphasize clinicians’ training to engage in these conversations. Two ways to accomplish this are by further exploring attitudes and perceptions of the veteran population within regional settings and gauging clinicians’ cultural competency in this domain to assess for biases or factors that limit candid exchanges between providers and patients. Further, with the expansion of extreme risk protection orders, there is a need to investigate their use in states that have enacted such policies and how these policies have impacted veteran suicide rates.

Our analysis must be interpreted in the context of several limitations. This study attempts to formalize and map the existing literature around firearm suicide among veterans. Additional search terms may have provided a broader scope of papers to review; however, we used a robust set of terms across a number of databases to capture the broadest set of the literature possible. The ability to make inferences is limited by the overwhelming cross-sectional nature and limited or very specific/homogenous sample populations of the included studies. Finally, since the completion of our systematic literature search and review, some additional papers on the topic have been published, which could provide additional context on this issue. However, because of the significant time period for publications included in this study, we are confident that our overview of this literature is robust. Although the review of the literature was thorough, because of the heterogeneity and low concordance of variables and findings across the studies, it was not feasible to perform a meta-analysis within the current body of the literature. For example, there were multiple different primary outcome measures across all studies, and different definitions of operationalized variables were used in the analyses; therefore, the study results are not easily compared with one another because of how those variables are defined across the studies.

CONCLUSIONS

The existing body of the literature on firearm suicide among veterans is preliminary, consisting of pilot studies and large cross-sectional studies of nonvalidated veterans using inconsistent definitions of veteran status. Heightened risk of suicide is well established, but this has not translated to improved screening for veterans within the VHA or other health systems. Limited data exist in non-VHA veteran populations, especially among women. Important next steps to strengthen our understanding of veteran firearm suicide include standardizing veteran status, implementing validation tools, and expanding the investigations to include service history components to aid with clinical risk stratification.

ACKNOWLEDGMENTS

None.

SUPPLEMENTARY MATERIAL

Supplementary material is available at Military Medicine online.

FUNDING

None declared.

CONFLICT OF INTEREST STATEMENT

None declared.

APPENDIX A

This consists of the search queries for PubMed, Embase, and PsycINFO.

Search Queries (February 21, 2019)

Embase

(“veteran”/exp OR veteran:ab,ti OR veterans:ti,ab OR “government”/exp OR “united states

department of veterans affairs”:ti,ab OR “veterans affairs”:ti,ab OR “veteran affairs”:ti,ab

OR “veterans health”/exp OR “veteran health”:ti,ab OR “veterans health”:ti,ab OR “military

family”/exp OR “military family”:ti,ab OR “military families”:ti,ab OR “families of military

personnel”:ti,ab OR “veterans families”:ti,ab OR “veteran families”:ti,ab OR “veterans

family”:ti,ab OR “soldier”/exp OR “military personnel”:ti,ab OR “military”:ti,ab OR “air force

personnel”:ti,ab OR “airman”:ti,ab OR “army personnel”:ti,ab OR “soldier”:ti,ab OR

“soldiers”:ti,ab OR “submariner”:ti,ab OR “submariners”:ti,ab OR “marine”:ti,ab OR

“marines”:ti,ab OR “navy personnel”:ti,ab OR “sailor”:ti,ab OR “sailors”:ti,ab OR “military

deployment”:ti,ab OR “combat deployment”:ti,ab OR “overseas deployment”:ti,ab OR “coast

guard”:ti,ab OR “coast guardsmen”:ti,ab OR “reservist”:ti,ab OR “national guard

personnel”:ti,ab OR “reserve personnel”:ti,ab OR “national guardsmen”:ti,ab)

AND

(suicide OR “suicide”:ti,ab OR “suicidal”:ti,ab OR “suicidal ideation”/exp OR “suicidal

ideation”:ti,ab OR “suicide ideation”:ti,ab OR “suicide attempt”/exp OR “suicide

attempt”:ti,ab OR “attempted suicide”:ti,ab)

AND

(“firearm”/exp OR “firearms”:ti,ab OR “rifle”:ti,ab OR “rifles”:ti,ab OR “shotgun”:ti,ab OR

“shotguns”:ti,ab OR “gun”:ti,ab)

PsycINFO

(DE “Military Veterans” OR veterans OR “veterans health” OR “veteran health” OR

“department of veterans affairs”OR “veterans affairs” OR “veteran affairs” OR “veterans

hospital” OR “veterans hospitals”OR (DE “Military Families” OR “Military Family” OR

“military family” or “military families”or “families of military personnel” OR “veterans

family” or “veterans families” or “veteran families” or “veterans family”) OR (DE

“Military Personnel” OR “Military Personnel” OR “military personnel” OR “military”

OR “air force personnel” OR “airman” OR “army personnel” OR “soldier” OR “soldiers”

OR “submariner” OR “submariners” OR “marine” OR “marines” OR “navy personnel”

OR “sailor” OR “sailors” OR “military deployment” OR “combat deployment” OR

“overseas deployment” OR “coast guard” OR “coast guardsmen” OR “reservist” OR

“national guard personnel” OR “reserve personnel” OR “national guardsmen”) OR DE

“Military Duty Status”)

AND

(DE “suicide” OR suicid* OR DE “Attempted Suicide” OR “suicide attempt” OR DE

“Suicidal Ideation” OR “suicide ideation”)

AND

(DE “Firearms” OR firearms OR gun OR shotgun OR rifle)

PubMed

(“veteran”[tiab] OR “veterans”[tiab] OR “veterans”[mh] OR “veterans health”[mh] OR

“veteran health”[tiab] OR “veterans health”[tiab] OR “United States Department of

Veterans Affairs”[Mesh] OR “veteransaffairs”[tiab] OR “veteran affairs”[tiab] OR

“Hospitals, Veterans”[Mesh] OR “veterans hospitals”[tiab]OR “veterans hospital”[tiab]

OR “Military Family”[mh] OR “military family”[tiab] or “military families”[tiab] or

“families of military personnel”[tiab] OR “veterans family”[tiab] or “veterans

families”[tiab] or “veteran families”[tiab] or “veterans family”[tiab] OR “Military

Personnel”[Mesh] or “military personnel”[tiab] or “military”[tiab] or “air force

personnel” [tiab] or “airman”[tiab] or “army personnel”[tiab] or “soldier”[tiab] or

“soldiers”[tiab] or “submariner”[tiab] or “submariners”[tiab] or “marine”[tiab] or

“marines”[tiab] or “navy personnel”[tiab] or “sailor”[tiab] or “sailors”[tiab] or “military

deployment”[tiab] or “combat deployment” [tiab] or “overseas deployment”[tiab] or

“coast guard”[tiab] or “coast guardsmen”[tiab] OR “reservist”[tiab] OR “national

guard personnel”[tiab] OR “reserve personnel”[tiab] OR “national guardsmen”[tiab])

AND

(“Suicide”[Mesh] OR “Suicide”[tiab] OR “Suicidal”[tiab] OR “Suicidal Ideation”[Mesh]

OR “suicidal ideation”[tiab]OR “suicide ideation”[tiab] OR “Suicide, Attempted”[Mesh]

OR “suicide attempt”[tiab] OR “attemptedsuicide”[tiab])

AND

(“firearms”[mh] OR “firearm”[tiab] OR “firearms”[tiab] OR “shotgun”[tiab] OR

shotguns[tiab] OR guns[tiab] OR gun[tiab] OR rifle[tiab] OR rifles[tiab])

APPENDIX B

Reviewed Papers’ Primary and Secondary Questions of Interest and Designs

TABLE B1.

Included Studies Published Between 1994 and 2019

First author, year (Ref. ID)	Study question	Study design
Anestis, 2016 (42)	(1) Is there an association between state-level veteran status and overall suicide rates?	Cross-sectional
	(2) Is there an association between state-level veteran status and firearm-specific suicide rates?
Balbuena, 2015 (33)	Is nicotine exposure an independent risk factor for suicide?	Case-control
Cleveland, 2017 (24)	(1) What is firearm ownership prevalence among veterans?	Cross-sectional (Survey)
	(2) Veteran attitudes about firearm ownership?
Desai, 2008 (41)	(1) What sociodemographic and clinical factors are associated with firearm suicide?	Cross-sectional
	(2) What state characteristics are associated with increased risk of suicide?
Dobscha, 2014 (38)	How do VHA clinicians react to positive suicide risk assessments in OIF/OEF veterans?	Cross-sectional
Freeman, 2003 (61)	(1) Is firearm ownership associated with psychiatric illness?	Cross-sectional
	(2) How do firearm-related behaviors differ among those with psychiatric illnesses?
Freeman, 1995 (62)	Is there a difference between veterans with PTSD who have a history of suicide attempt compared with those who do not?	Cross-sectional
Heinz, 2016 (46)	What is the association between veterans with psychological and behavioral suicide risk factors and firearm ownership?	Cross-sectional (Survey)
Hoffmire, 2014 (29)	What is the association between veteran status and firearm suicide?	Cross-sectional
Kaplan, 2009 (30)	Do events before a suicide differ between veterans who use a firearm compared with those who use another method?	Cross-sectional
Kaplan, 2009 (25)	What is the prevalence of firearm suicide among veterans compared with nonveterans?	Cross-sectional (Descriptive)
Kaplan, 2012 (40)	What associations or events in the time leading up to a firearm suicide among men 65 years or older?	Cross-sectional
Kaplan, 2012 (32)	Are veterans at higher risk of suicide than nonveterans?	Cross-sectional
Kaplan, 2012 (58)	What is the association between alcohol use and suicidal behaviors?	Cross-sectional
Liu, 2013 (59)	What are the sociodemographic factors associated with firearm suicide?	Cross-sectional
Massetti, 2018 (39)	How do events leading up to a suicide differ between patients with and without a cancer diagnosis?	Case-control (Matched)
McCarten, 2015 (31)	How does the prevalence of veteran firearm suicide change between 2001 and 2010?	Cross-sectional (Time series)
McCarthy, 2012 (17)	Do suicide risk factors, methods, and rates differ between VHA-users in rural compared with urban settings?	Cohort study (Prospective)
McGlade, 2016 (48)	In what ways do veteran and nonveteran suicide decedents differ in Utah?	Cross-sectional
Miller, 2009 (21)	Are veterans at increased risk of suicide compared with those in the general population?	Cohort study (Prospective)
Miller, 2012 (50)	(1) Are previous estimates of increase suicide risk among veterans compared with the general population reproducible?	Cross-sectional
	(2) Is there an association between veteran status and firearm suicide?
	(3) Is veteran status a risk factor for suicide?

TABLE B1.

(Continued)

Seyfried, 2011 (34)	Are there predictors for suicide among patients with dementia?	Cohort study (Retrospective)
Simonetti, 2018 (22)	What are firearm storage practices among veteran firearm owners?	Cross-sectional (Survey)
Simonetti, 2018 (23)	(1) How do veterans store their firearms?	Cross-sectional (Survey)
	(2) Do veterans associate firearm access with suicide risk?
	(3) Would veterans change their firearm storage habits if household member was at increased risk?
Smith, 2015 (47)	(1) Among VHA residential PTSD patients, how many own firearms?	Cross-sectional (Survey)
	(2) Is there an association between firearm ownership and suicidal thoughts and behavior in veterans with PTSD?
	(3) How do veteran firearm owners with PTSD store their firearms?
Strom, 2012 (63)	What is the prevalence of risky behavior among veterans with PTSD?	Cross-sectional (Survey)
Swanson, 2018 (60)	(1) Is there evidence to support restricting firearm rights because of fiduciary assignment?	Case-control
	(2) Is there a relationship between financial competency and risk of harm to others or self?
Thompson, 2006 (37)	What is the prevalence of suicidal ideation among veterans being treated for substance use?	Cross-sectional (Survey)
Tripp, 2016 (43)	(1) Is previous wartime experience in Iraq or Afghanistan associated with suicidal ideation?	Cross-sectional (Survey)
Trivedi, 2017 (35)	(1) What is life expectancy of veterans with mental illness?	Cross-sectional (Time series)
	(2) What are the most common causes of death among veterans with psychiatric illness?
	(3) Is there a difference in life expectancy when comparing veterans with psychiatric illness to those without?
	(4) What are the VHA utilization patterns among veterans?
Valenstein, 2018 (45)	Patients’ attitudes toward voluntary firearm restriction interventions for DVA mental health and substance use patients?	Cross-sectional (Survey)
Valenstein, 2019 (44)	(1) Veteran attitudes about firearm ownership?	Cross-sectional (Survey)
	(2) Attitudes toward receiving clinical firearm safety counseling?
	(3) Frequency firearm safety discussed by clinician?
Walters, 2012 (49)	(1) What are DVA stakeholders’ attitudes about firearm safety among VHA patients?	Qualitative study
	(2) What firearm safety interventions are VHA patients and stakeholders receptive to?
Coyne-Beasley, 2010 (36)	What is the breakdown of violent deaths among veterans in North Carolina?	Cross-sectional
Sinha, 2018 (64)	What data exist for risk assessments in veteran suicide?	Case study
Warren, 2010 (65)	What were the characteristics of fatal suicide among patients at a hospice facility?	Case study
Freeman, 1994 (26)	What are the characteristics of firearm ownership and use among combat veterans with PTSD?	Cross-sectional

Abbreviations: OEF, Operation Enduring Freedom; OIF, Operations Iraqi Freedom; VHA, Veterans Health Administration.

STUDY QUESTIONS AND DESIGN

Summary of study questions: Importantly, because of the predominant cross-sectional nature of these studies, the authors interrogated datasets such that several questions could be asked within each study. We listed the points that focused on statistically significant findings and other findings that are relevant for clinical approaches to firearm suicide risk reduction.

Author notes

An early version of this manuscript was presented at an online venue on May 8, 2019 at the Johns Hopkins Bloomberg School of Public Health Capstone Symposium.