Abstract
Introduction:
There are meager individual-level data on long-term predictors of firearm suicide.
Methods:
This was an analysis of males (n=189,558) in the Project Talent (PT) cohort, a national probability sample of high school schools in 1960 when students completed a baseline PT self-report inventory. Mortality follow-up was contingent on survival until 1979, the onset of the National Death Index, when the cohort was mean age 35.7. Mortality follow-up continued until death or age 75, reached by all surviving members by 2018. Analyses were conducted in 2022, with the main outcome firearm suicide deaths (n=479). Factor analyses of PT items yielded three key factors: (1) interests in firearm-related professions (i.e., military service, police force); (2) interests in hunting or fishing and knowledge of long guns; and (3) stereotypic masculinity.
Results:
Survival analyses showed long-term risk for firearm suicide was associated with 1-standard–deviation increases in firearm-related vocational interests in adolescence, adjusted hazard ratio, AHR [95% CI] = 1.23 [1.09, 1.40], and masculinity, AHR [95% CI] = 1.15 [1.04, 1.28]. Decreased long-term firearm suicide risk was associated with increased hunting interests and knowledge of long guns in adolescence, AHR [95% CI] = 0.86 [0.77, 0.96] and competitive sports participation, an exploratory variable, AHR [95% CI] = 0.89 [0.80, 0.99].
Conclusions:
Prevention efforts are needed to lower long-term firearm suicide risk among adolescent males with high stereotypic masculinity and those interested in military or police service. Potential protective effects of competitive sports participation and socialization to long guns through hunting require further study.
Introduction
After modest drops in suicide rates during the COVID-19 pandemic, the suicide rate in the U.S. rose sharply in 2021, reaching an all-time high1. The increased rate is a continuation of a longstanding trend, with rates increasing by more than one-third since 2000, with greater increases in select populations including Black male youth.2–3 Approximately 78% of suicide decedents are males, 58% use a firearm, and males account for 87% of all firearm suicide deaths1, underscoring the importance of research and prevention of firearm suicide in US males. Indeed, firearms are the most deadly method of self-harm4; firearm suicide risk is higher in demographic groups with high firearm ownership (e.g., rural men),5 and access to a firearm, particularly a handgun, is a potent risk factor.6
A life course perspective stresses the importance of examining earlier periods of life to ascertain their downstream effects on health outcomes7, with adolescence being viewed as a key period for health outcomes far into the future,8 including suicide.9 In the current study, long-term risk for firearm suicide associated with firearm-related socialization in adolescent males was examined. Firearm-related socialization is defined as the process of acquiring values, behaviors, and attitudes related to firearm ownership and use. There has been illuminating ecological-level firearm-related socialization research,10,11 but there is a lack of individual-level studies linking firearm-related socialization in adolescence with firearm injury deaths in adulthood, the gap addressed herein.
Social cognitive theory posits that behaviors and attitudes are learned vicariously through observation, modeling, and reinforcement of behavior (approval, praise, punishment), including through media.12 Adolescents’ identification with their parents and peers; inclinations for status, affection, and avoidance of rejection; and exposure to a behavior or attitude serve to make it more attractive.13 Research on adolescents shows that personal experience with firearms (e.g., hunting), household ownership of firearms, peer use of firearms, and positive family and peer attitudes toward firearms influence personal attitudes toward firearms and inclinations toward their ownership and use.10,14 Some research also reports associations of suicidal thoughts and behavior in adults with access to a gun in childhood15 and prior experience firing a gun.16
Socialization and learning are influenced by personality characteristics present in adolescence; for example, neuroticism, or its opposite, “calm,” which can also influence long-term health behaviors and outcomes, including all-cause mortality.17 However, limited data exist on personality variables measured in youth and long-term risk for suicide. In the study of firearm suicide, personality features associated with firearm interests and ownership warrant special attention, including stereotypic masculinity.18 One report showed that stereotypic masculinity measured in adolescence was associated with prospective risk for suicide deaths in US males.19 All but one of the suicide deaths were carried out with a firearm, suggesting a link between masculinity and risk for firearm suicide per se, but the small number of suicide deaths in the cohort (n=22) makes this idea speculative. In contrast, a report of Swedish military conscripts showed that lower stereotypic masculinity increased risk for suicide20, but the analyses did not examine firearm suicide per se, and generalizability to the U.S. is unclear.
A large cohort of US males who were first examined in high school were linked with mortality records over four decades in the life course.9 The primary objective of analysis was to test the hypothesis that firearm-related socialization in adolescence, consisting of contextual variables that indicate exposure to firearms (e.g., practical knowledge about firearms) and that signal an adolescent’s interest in firearm-related vocations (e.g., desire to enlist in armed forces) confer long-term risk for firearm suicide in US males. The secondary objective was to test the hypothesis that stereotypic masculinity in adolescence confers long-term risk for firearm suicide.
Methods
Study Population
Project Talent (PT) began in 1960 as a national probability sample of 5% of all US high schools (n = 1,226). All 9th to 12th grade students (n = 377,016) in these schools completed a range of tests and questionnaires of cognitive skills, academic achievement, hobbies and interests, and personal and family life. The current analysis is restricted to males in the cohort (n=189,558). Specifically, males in PT were followed beginning in 1979, coinciding with the onset of the National Death Index (NDI), when the cohort was mean (standard deviation) age 35.7 (1.3) years. Males in PT were followed until age 75, when the data were censored, with all surviving cohort members reaching age 75 by the end of 2018. For ease of communication, we refer to cohort members who were not identified as deceased beginning in 1979 as ‘survivors’, but this is not a count of confirmed survivors because the procedures missed deaths that occurred earlier in the life course.
Participants were originally tracked using commercial databases, including LexisNexis, along with web sources including Ancestry.com. A subset of the sample was also linked to health records from the Centers for Medicare and Medicaid Services. These sources, in combination with the NDI, were used to identify PT sample members who were deceased. The NDI was also used to determine cause of death in deceased members of the cohort. See Appendix 1 for detailed description of the methods used to integrate these sources of information and promote quality control. Informed consent regulations were not established in 1960: therefore, informed consent was not obtained at the base year data collection. The study was approved by Institutional Review Boards at the University of Rochester Medical Center and the American Institutes for Research.
Measures
Firearm-related socialization and interest scores were the primary predictors. Ten relevant items in the baseline PT survey were identified: 3 items related to knowledge of firearms—specifically, long guns (e.g., what “double action” refers to); 5 items indicating interests in vocations where firearms are used (e.g., interest in becoming a police officer); and 2 items assessing current hobbies/activities related to hunting or fishing. The baseline PT sample was randomly split into exploratory and confirmatory halves. In the exploratory subsample, initial exploratory factor analysis determined the number of factors, and then exploratory structural equation modeling was used to determine initial fit and modifications needed to achieve adequate fit.21 The exploratory factor analysis produced a scree plot indicating a 2-factor solution, with 1 factor corresponding primarily to hunting/fishing interests and firearm knowledge and the other to firearm-related vocational interests. A confirmatory factor analysis showed this 2-factor model produced excellent fit (comparative fit index = .97, root mean square error of approximation of .052), and nearly identical fit statistics were obtained in the validation subsample. The 2 factors were correlated at .56. A full information maximum likelihood model was then fit in the full sample to estimate factor scores for those with at least two-thirds of the items (96%). Table 1 shows the loadings from this 2-factor model. Marked sex differences in the factor scores were noted, consistent with the era when the data were collected, that were as large as the measures of masculinity and femininity (described next). These results validated the decision to focus analyses of firearm-related socialization and interests and suicide in males. Z-scores within sex were created, with a 1-unit change in scores indicating 1 standard deviation among males. See Appendix 2 for more information.
Table 1.
Firearm-related socialization and interest factors
| Item | F1 | F2 |
|---|---|---|
| Hunting /fishing is a hobby | .70 | |
| Interested in being a professional hunter | .80 | |
| Knows which of several guns has the largest bore | .44 | |
| Knows what double action refers to | .02 | |
| Knows how many shells in box of .22 caliber ammo | .09 | |
| Military drills is a hobby | .68 | |
| Interested in becoming police officer | .53 | |
| Interested in becoming army officer | .77 | |
| Interested in becoming marine officer | .83 | |
| Interested in becoming air force officer | .82 |
Notes. Scores reflect item loadings on each factor. Factor 1 (F1) = hunting/fishing interests and firearm knowledge. Factor 2 (F2) = firearm-related vocational interests.
Masculinity, femininity, and competitive sports scores were secondary predictors of interest. Coleman and colleagues’ definition of “traditional” masculinity—“a set of norms that includes competitiveness, emotional restriction, and aggression”19—was adopted. The measurement scope was expanded to include stereotypic femininity, consistent with work treating masculinity and femininity as separate dimensions.22 All items in the baseline survey were identified that expressed highly gender-stereotyped activities (e.g., metal working, sewing), job aspirations (e.g., bricklayer, secretary), knowledge (e.g., sports, home economics), or personality traits (e.g., sensitivity)—a process that led to the identification of 27 candidate items. The preliminary exploratory factor analysis in the development sample yielded a scree plot suggesting 3 factors that explained 97% of the common variance. This solution held up in the confirmatory factor analysis. Factor 1 captured traditionally masculine interests and activities, factor 2 traditionally feminine interests and activities, and factor 3 competitive sports. Factors 1 and 2 were correlated at −.47, factors 1 and 3 were correlated at .39, and factors 2 and 3 were correlated at −.42. A test of the criterion validity of these scores is their difference between males and females. Factor scores were standardized by the overall sample standard deviation for ease of interpretation and to examine sex differences; raw scores yielded similar results. The distribution of z-scores reflected marked sex differences: mean z-scores for males/females for stereotypic masculinity were .67/–.67, for stereotypic femininity .84/–.85, and for competitive sports .58/–.57, with all comparisons highly significant (P < .001). Consistent with the focus on males, z-scores within sex for masculinity, femininity, and competitive sports interest/activities were created. See Appendix 1 for more information.
To assess outcomes, deaths were coded as suicide if their underlying cause corresponded to ICD-9 (pre-1998) codes of 950.XX-959.XX, and ICD-10 codes of X60–84, Y87.0. Firearm-related suicides were defined based on ICD-9 codes 955 and 955.4 and ICD-10 codes X72–74. These deaths represent suicide codes for which intent is clear. Unintentional poisonings and deaths with unclear intent were not coded as suicides.
Covariates including race/ethnicity, rurality, census region, and family of origin socioeconomic status (SES) were included in statistical models because of the salience of these variables to firearm suicide risk in US men.5,9,23 Personality measures of neuroticism and impulsivity were also covaried because of their salience in suicidal behavior.24 Specifically, personality was measured at baseline by the PT Personality Inventory, which assessed 10 personality constructs—notably, “calm” and “impulsivity”—and has since been mapped to the five-factor model traits including neuroticism.25 Higher scores on the calm scale are strongly inversely related to neuroticism.25 Therefore, the inverse of calm was used to represent neuroticism.
Statistical analysis
The analyses were conducted in 2022. The models are contingent on surviving until 1979 and included transitions from alive to (1) firearm suicide, (2) non-firearm suicide, and (3) all other forms of non-suicide–related mortality. The last category was included to account for censoring because of competing risks. All predictors were examined for violations of the proportional hazards’ assumption, and for linearity in the log hazard (i.e., whether predictors showed quadratic or other non-linear associations with the outcome on the log hazard scale, which would necessitate transformation of those predictors to achieve proper model specification). The cubic-spline based analysis of best-fitting distribution described in Crowther & Lambert26 was conducted and, based on it, the Weibull distribution was used in analyses because it afforded the best fit (i.e., based on the Bayesian Information Criterion and Akaike Information Criterion). The multistate models jointly modeled the primary outcome and competing risk.26–27 These models used attained age as the time scale, with entry being age at baseline (i.e., the computation of time to event is the attained age at death or censoring, minus the age at baseline).28 An initial set of models examined the firearm socialization and masculinity factors one at a time. A final model included all predictors simultaneously, including covariates.
Results
There were 152,501 male survivors in the PT cohort through age 75, along with 686 suicide deaths (479 firearm, 207 non-firearm) and 36,371 deaths by other causes which are not a focus of this report. Among firearm suicide decedents, ages at death were as follows: range 35–75 years, median age 66, 25th percentile 55 years, 75th percentile 71 years. Among non-firearm suicide decedents, ages at death were as follows: range 35–75 years, median age 59.5, 25th percentile 48 years, 75th percentile 68 years. See Figure 1 for flow of PT cohort members. Table 2 shows descriptive statistics measured in adolescence for male survivors, firearm suicide decedents, and non-firearm suicide decedents. Most members of the cohort were white and non-Hispanic (~91.1%), most resided in a non-rural area (~87.6%), and they were dispersed throughout the four regions of the US.
Figure 1 :
Flow of Project Talent Cohort
1The authors have knowledge of only 9 individuals who died, with no information on cause of death, in the cohort of Project Talent males prior to 1979.
2Mortality follow-up began in 1979, with the onset of the National Death Index (NDI), and continued until death or age 75, an age reached by all surviving members by 2018.
Table 2.
Project Talent male sample descriptive statistics
| Standardized Predictors Assessed in Adolescence | Survivors Through Age 75 (N = 152,501) | Non-Firearm Suicide Decedents (N = 208) | Firearm Suicide Decedents (N = 466) |
|---|---|---|---|
| White, Non-Hispanic | 138,985 (91.1%) | 191 (91.8%) | 443 (95.1%) |
| Rural Resident | 18,840 (12.4%) | 15 (7.2%) | 55 (11.8%) |
| Census Region 1 | 40,658 (26.7%) | 75 (36.1%) | 102 (21.9%) |
| Census Region 2 | 49,335 (32.4%) | 54 (36.0%) | 141 (30.3%) |
| Census Region 3 | 43,108 (28.3%) | 49 (23.6%) | 154 (33.1%) |
| Census Region 4 | 19,400 (12.7%) | 30 (14.4%) | 69 (14.8%) |
| Family of Origin Socio-Economic Status | 02 (1) | .29 (.95) | .02 (.95) |
| Vocational Interests (in police, branch of military) | 00 (1) | −.26 (1.10) | .15 (.93) |
| Hunting/fishing Interests and Firearm Knowledge | −01 (1) | −.22 (1.10) | −.07 (1.03) |
| Masculinity | −02 (1) | −.27 (1) | .12 (.99) |
| Femininity | −.02 (.99) | −.01 (1) | −.15 (.90) |
| Competitive Sports Participation | 01 (1) | −.05 (1.06) | −.06 (1.01) |
| Calm | .01 (1) | .00 (1.01) | −.09 (.97) |
| Impulsivity | −.01 (1) | .02 (.99) | −.01 (1.02) |
Notes: Mean (standard deviation) values shown, except for categorical variables (top 3 rows). Predictors standardized (mean = 0, standard deviation = 1) within Project Talent male participants. Census regions: 1=northeast, 2=midwest, 3=south, 4=west.
Table 3 shows the parameter estimates for each outcome of interest (i.e., firearm, non-firearm suicide). Each predictor of interest was associated with firearm suicide (P < .05). A 1 standard deviation increase in firearm-related vocational interests in adolescence was associated with increased risk for firearm suicide (adjusted hazard ratio, AHR [95% CI] = 1.23 [1.09, 1.40], whereas an increase in hunting/fishing interests and long gun firearm knowledge was associated with lowered risk for firearm suicide, AHR [95% CI] = 0.86 [0.77, 0.96]. An increase of 1 standard deviation in masculinity in adolescence was associated with increased risk for firearm suicide, AHR [95% CI] = 1.15 [1.04, 1.28], whereas an increase in femininity was associated with lowered risk for firearm suicide, AHR [95% CI] = 0.89 [0.79, 0.99]. An increase of 1 standard deviation in competitive sports participation in adolescence, an exploratory variable, was associated with decreased risk for firearm suicide, AHR [95% CI] = 0.89 [0.80, 0.99]. Among covariates, white, non-Hispanic males and those who resided in the South or in a rural area in adolescence were at increased risk for firearm suicide.
Table 3.
Predictors of firearm and non-firearm suicide
| Predictor – Measured in Adolescence | Non-firearm suicide HR (95% CI) | Firearm suicide HR (95% CI) | Non-firearm suicide AHR (95% CI) | Firearm suicide AHR (95% CI) |
|---|---|---|---|---|
| White, Non-Hispanic (vs. Minority) | 1.12 (0.68, 1.83) | 1.92 (1.26, 2.92) | 0.98 (0.56, 1.74) | 1.60 (1.01, 2.53) |
| Rural Residence (vs. Non-Rural) | 1.83 (1.08, 3.10) | 1.06 (0.80, 1.41) | 1.27 (1.12, 1.43) | 1.58 (1.48, 1.67) |
| Region 2 (vs. 1) | 0.59 (0.41, 0.83) | 1.12 (0.87, 1.45) | 0.63 (0.44, 0.91) | 1.04 (0.80, 1.37) |
| Region 3 (vs. 1) | 0.60 (0.42, 0.86) | 1.39 (1.08, 1.79) | 0.69 (0.47, 1.02) | 1.38 (1.05, 1.81) |
| Region 4 (vs. 1) | 0.84 (0.55, 1.28) | 1.41 (1.04, 1.92) | 0.84 (0.52, 1.34) | 1.35 (0.97, 1.88) |
| Family of Origin SES | 1.33 (1.16, 1.54) | 1.01 (0.92, 1.11) | 1.27 (1.08, 1.50) | 1.04 (0.93, 1.15) |
| Vocational Interests (in police, military branch) | 0.79 (0.69, 0.89) | 1.17 (1.06, 1.30) | 0.87 (0.74, 1.03) | 1.23 (1.09, 1.40) |
| Hunt/Fish Interests and Firearm Knowledge | 0.81 (0.71, 0.93) | 0.94 (0.85, 1.03) | 0.96 (0.81, 1.14) | 0.86 (0.77, 0.96) |
| Masculinity | 0.76 (0.66, 0.88) | 1.13 (1.03, 1.25) | 0.86 (0.73, 1.01) | 1.15 (1.04, 1.28) |
| Femininity | 1.00 (0.87, 1.15) | 0.86 (0.78, 0.95) | 1.03 (0.88, 1.20) | 0.89 (0.79, 0.99) |
| Competitive Sports Participation | 0.95 (0.82, 1.09) | 0.93 (0.85, 1.03) | 0.98 (0.84, 1.14) | 0.89 (0.80, 0.99) |
| Calm | 1.00 (0.87, 1.15) | 0.91 (0.83, 1.00) | 0.95 (0.82, 1.11) | 0.93 (0.84, 1.03) |
| Impulsivity | 1.02 (0.89, 1.17) | 0.99 (0.91, 1.09) | 1.01 (0.87, 1.17) | 1.01 (0.91, 1.11) |
Notes. HR = unadjusted hazard ratio. AHR = fully adjusted hazard ratio. Predictors standardized within PT male participants, except for categorical variables (top 3 rows). Census regions: 1=northeast, 2=midwest, 3=south, 4=west. Boldface indicates statistical significance, p<0.05.
None of the predictors of interest were associated with non-firearm suicide at a statistically significant level (P < .05). Residing in the Midwest in adolescence was associated with decreased risk for non-firearm suicide, whereas residing in a rural area and with a family of origin with greater SES were associated with increased risk.
Discussion
The study examined firearm-related socialization and interest and masculinity measured in high school with risk for firearm suicide over much of the adult life course in US males. Specifically, suicides between ages 35 and 75 were examined, a period in life when approximately 57% of suicides in US males occur.1 As hypothesized, interest in service in the military and police forces in adolescence are associated with increased risk for firearm suicide in US men. The extent that these vocational interests in adolescence actually led to service in these professions which are associated with increased firearm suicide risk5,26 may explain the finding. Mechanisms may include further firearm-related socialization through these vocations, increased access to firearms through their vocational role, increased likelihood of personal ownership of a firearm among police and military service members, and the potential role of firearm use in ‘acquired capability for suicide’, a theorized psychological risk factor.16,30 As hypothesized, stereotypic masculinity in adolescence was associated with increased risk for firearm suicide. Associations of masculinity with greater interest and ownership of firearms may be a mechanism.18 Another result was that femininity in adolescence lowered risk for firearm suicide, a logical result insofar as it is in the opposite direction of the result obtained for masculinity.
Hunting/fishing interests and firearm knowledge in adolescence lowered risk for firearm suicide, a result that was not hypothesized. Although the role of long guns in suicide in the US should not be dismissed,31 handguns are used in more than three-quarters of firearm suicides.31 Yet, knowledge about long guns was the focus of the PT firearm knowledge items. A possible mechanism for a protective effect is that socialization to long guns, including through hunter education courses which emphasize safety to humans, creates the expectation that firearms are to be used exclusively in hunting or sport, lowering the potential for diversion for harmful use towards humans, including in suicide. However, quality research to test this idea are lacking, and the diversity of hunter education courses and licensing requirements which vary by state, create significant challenges to research efforts.
Competitive sports participation in adolescence, an exploratory variable, was also associated with lowered risk for firearm suicide, consistent with data on mental health benefits of sports participation in youth, benefits that may be attributable to the health promotion effects of physical exercise, increased connectedness with peers and adult role models (e.g., coaches), among other mechanisms.32 Results pertaining to the covariates were unsurprising with the exception of the result indicating increased risk for non-firearm suicide associated with higher family of origin SES. Further research on this question is needed.
Limitations
The study did not examine firearm suicide in women.33 The cohort was gathered in 1960 when smaller portions of the U.S. population were racial and ethnic minorities, and when there were more rigid gender roles and expectations. There is unclear generalizability of results to non-high school attendees. Mortality follow-up did not begin until 1979, when the cohort was mean age 35.7 years, and it did not extend beyond age 75. Among US males, approximately 33% of suicides occur before age 35 and 10% occur after age 751, suicides that would be missed in the analyses. Assessments of variables occurring between adolescence and death including vocational, medical, social, behavioral, and environmental factors were not available. Of particular relevance, PT cohort members born in 1944 or later were eligible for the military draft, representing 52.5% of the sample, yet data on military service including deployment to Vietnam, either as conscripts or enlistees, were not available. Likewise, data on service in police professions were not available. Measures of firearm-related socialization and masculinity were developed within the sample. PT knowledge items referred to long guns used in hunting. Because socialization to hunting involves these guns, the relevance to privately owned assault weapons which have become more common is unclear. Interests in hunting or fishing were assessed (rather than interests in hunting alone). Region, rurality, or other macro-level variables associated with firearm suicide at the time of death were not considered because these were not available in survivors in the cohort over follow-up.
Conclusions
A range of approaches is required to prevent firearm suicide deaths.34 Interests in military and police vocations and stereotypic masculinity are common characteristics in adolescence that predict firearm suicide risk over much of the adult life course in men, underscoring the practical significance and public health importance of developing preventive strategies for males with these characteristics. Results further suggest competitive sports participation in adolescence as well as education and socialization to long guns through hunting and related interests decrease long-term firearm suicide risk. However, these potential protective effects are novel and were not hypothesized, requiring further research.
Supplementary Material
Acknowledgements
This work was supported by award 1R61AG072408-01 (Chapman, Conner) from the National Institute on Aging, National Institutes of Health. The results have not been previously reported elsewhere. The research presented in this paper is that of the authors and does not reflect the official policy of the NIH. The study sponsor had no role in the study design; collection, analysis, and interpretation of data; writing the report; or the decision to submit the report for publication. No financial disclosures have been reported by the authors of this paper.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.Centers for Disease Control and Prevention. WISQARS (Web-based Injury Statistics Query and Reporting System). CDC.gov. Published 2021. https://www.cdc.gov/injury/wisqars/index.html. Accessed September 16, 2023.
- 2.Garnett MF, Curtin SC, Stone DM. Suicide mortality in the United States, 2000–2020. NCHS Data Brief; 2022;433. 10.15620/cdc:114217 [DOI] [PubMed] [Google Scholar]
- 3.Hedegaard H, Curtin SC, Warner M. Suicide mortality in the United States, 1999–2019. NCHS Data Brief; 2021;398. 10.15620/cdc:101761. [DOI] [PubMed] [Google Scholar]
- 4.Conner A, Azrael D, Miller M. Suicide case-fatality rates in the United States, 2007 to 2014: a nationwide population-based study. Ann Intern Med. 2019;171(12):885–895. doi: 10.7326/M19-1324 [DOI] [PubMed] [Google Scholar]
- 5.Olfson M, Cosgrove CM, Wall MM, Blanco C. Sociodemographic factors associated with high risk for firearm suicide among US adults. JAMA Intern Med. 2020;180(7):1014–1019. doi: 10.1001/jamainternmed.2020.1334 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Studdert DM, Zhang Y, Swanson SA, et al. Handgun ownership and suicide in California. N Engl J Med. 2020;382(23):2220–2229. doi: 10.1056/NEJMsa1916744 [DOI] [PubMed] [Google Scholar]
- 7.Kuh D, Ben-Shlomo Y, Lynch J, Hallqvist J, Power C. Life course epidemiology. J Epidemiol Community Health. 2003;57(10):778–83. doi: 10.1136/jech.57.10.778 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Viner RM, Ross D, Hardy R, et al. Life course epidemiology: recognising the importance of adolescence. J Epidemiol Community Health. 2015;69(8):719–720. doi: 10.1136/jech-2014-205300 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Chapman BP, Peters K, Kannan V, et al. Intergenerational social mobility and suicide risk by firearm and other means in US males. J Psychiatr Res. 2022;156:318–323. doi: 10.1016/j.jpsychires.2022.10.037 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Lizotte AJ, Tesoriero JM, Thornberry TP, Krohn MD. Patterns of adolescent firearms ownership and use. Justice Q. 1994;11(1):51–74. [Google Scholar]
- 11.Wombacher C, Wallace LN. Childhood gun socialization among college-educated youth: shaping adult attitudes. Soc Sci J. 2019;56(3):349–357. doi: 10.1016/j.soscij.2018.08.005 [DOI] [Google Scholar]
- 12.Bandura A Social cognitive theory: an agentic perspective. Annu Rev Psychol. 2001;52:1–26. doi: 10.1146/annurev.psych.52.1.1 [DOI] [PubMed] [Google Scholar]
- 13.Telzer EH, van Hoorn J, Rogers CR, Do KT. Social influence on positive youth development: a developmental neuroscience perspective. Adv Child Dev Behav. 2018;54:215–258. doi: 10.1016/bs.acdb.2017.10.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Ding CS, Nelsen EA, Lassonde CT. Correlates of gun involvement and aggressiveness among adolescents. Youth Soc. 2002;34(2):195–213. [Google Scholar]
- 15.Copeland WE, Tong G, Gifford EJ. Childhood gun access, adult suicidality, and crime. Pediatrics. 2021;148(2):e2020042291. doi: 10.1542/peds.2020-042291 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Anestis MD, Capron DW. Deadly experience: the association between firing a gun and various aspects of suicide risk. Suicide Life Threat Behav. 2017;48(6):699–708. doi: 10.1111/sltb.12381 [DOI] [PubMed] [Google Scholar]
- 17.Chapman BP, Huang A, Horner E, et al. High school personality traits and 48-year all-cause mortality risk: results from a national sample of 26 845 baby boomers. J Epidemiol Community Health. 2019;73(2):106–110. doi: 10.1136/jech-2018-211076 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Borgogna NC, McDermott RC, Brasil KM. The precarious masculinity of firearm ownership. Psychol Men Masc. 2021;23(2):173–182. doi: 10.1037/men0000386 [DOI] [Google Scholar]
- 19.Coleman D, Feigelman W, Rosen Z. Association of high traditional masculinity and risk of suicide death: secondary analysis of the Add Health study. JAMA Psychiatry. 2020;77(4):435–437. doi: 10.1001/jamapsychiatry.2019.4702 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Månsdotter A, Lundin A, Falkstedt D, et al. The association between masculinity rank and mortality patterns: a prospective study based on the Swedish 1969 conscript cohort. J Epidemiol Community Health. 2009;63(5):408–413. doi: 10.1136/jech.2008.082628. [DOI] [PubMed] [Google Scholar]
- 21.Asparouhov T, Muthén B. Exploratory structural equation modeling. Struct Equ Modeling. 2009;16(3):397–438. 10.1080/10705510903008204 [DOI] [Google Scholar]
- 22.Choi N, Fuqua DR. The structure of the Bem Sex Role Inventory: a summary report of 23 validation studies. Educ Psychol Meas. 2003;63(5):872–887. 10.1177/0013164403258235 [DOI] [Google Scholar]
- 23.Rees CA, Monuteaux MC, Steidley I, et al. Trends and disparities in firearm fatalities in the United States, 1990–2021. JAMA Netw Open. 2022;5(11):e2244221. doi: 10.1001/jamanetworkopen.2022.44221 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Moore FR, Doughty H, Neumann T, McClelland H, Allott C, O’Connor RC. Impulsivity, aggression, and suicidality relationship in adults: a systematic review and meta-analysis. EClinicalMedicine. 2022;45:101307. doi: 10.1016/j.eclinm.2022.101307 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Pozzebon J, Damian RI, Hill PL, Lin Y, Lapham S, Roberts BW. Establishing the validity and reliability of the Project Talent Personality Inventory. Front Psychol. 2013;4:968. 10.3389/fpsyg.2013.00968 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Crowther MJ, Lambert PC. Parametric multistate survival models: Flexible modelling allowing transition‐specific distributions with application to estimating clinically useful measures of effect differences. Stat Med. 2017; 36(29):4719–4742. doi: 10.1002/sim.7448 [DOI] [PubMed] [Google Scholar]
- 27.Royston P, Lambert PC. Flexible parametric survival analysis using Stata: Beyond the Cox model. College Station, TX: Stata Press, 2011. [Google Scholar]
- 28.Thiébaut ACM, Bénichou J Choice of time‐scale in Cox’s model analysis of epidemiologic cohort data: a simulation study. Stat Med. 2004; 23(24):3803–3820. doi: 10.1002/sim.2098. [DOI] [PubMed] [Google Scholar]
- 29.Violanti JM, Steege A. Law enforcement worker suicide: an updated national assessment. Policing. 2021;44(1):18–31. doi: 10.1108/PIJPSM-09-2019-0157 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Smith PN, Cukrowicz KC. Capable of suicide: a functional model of the acquired capability component of the Interpersonal-Psychological Theory of Suicide. Suicide Life Threat Behav. 2010;40(3):266–75. doi: 10.1521/suli.2010.40.3.266. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Hanlon TJ, Barber C, Azrael D, Miller M. Type of firearm used in suicides: findings from 13 states in the National Violent Death Reporting System, 2005–2015. J Adolesc Health. 2019;65(3):366–370. doi: 10.1016/j.jadohealth.2019.03.015 [DOI] [PubMed] [Google Scholar]
- 32.Eime RM, Young JA, Harvey JT, Charity MJ, Payne WR. A systematic review of the psychological and social benefits of participation in sport for children and adolescents: informing development of a conceptual model of health through sport. Int J Behav Nutr Phys Act. 2013;10(1):1–21. 10.1186/1479-5868-10-98 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Miller M, Zhang Y, Prince L, et al. Suicide deaths among women in California living with handgun owners vs those living with other adults in handgun-free homes, 2004–2016. JAMA Psychiatry. 2022;79(6):582–588. doi: 10.1001/jamapsychiatry.2022.0793 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Betz ME, Thomas AC, Simonetti JA. Firearms, suicide, and approaches for prevention. JAMA. 2022;328(12):1179–1180. doi: 10.1001/jama.2022.16663. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.