Childhood Gun Access, Adult Suicidality, and Crime

A completed Strengthening the Reporting of Observational Studies in Epidemiology checklist is provided. The Great Smoky Mountains Study (GSMS) is a longitudinal, representative study of children in 11 predominantly rural counties of North Carolina.¹⁸ . Three cohorts of children, ages 9, 11, and 13 years, were recruited from a pool of some 12 000 children by using a 2-stage sampling design, resulting in N = 1420 participants (49% female; see also for more details¹⁸ ). First, potential participants were randomly selected from the population by using a household equal probability design. Next, participants were screened for risk of psychopathology; participants screening high were oversampled in addition to a random sample of the rest. In addition, American Indians were oversampled to constitute 25% of the sample. Sampling weights inversely proportional to participants’ probability of selection allows representative prevalence estimates despite oversampling. See Supplemental Figure 2 and references^14,18,19  for additional detail.

Annual assessments were completed on the 1420 children until age 16 (6674 observations of 1420 individuals; 1993 to 2000) and then again at ages 19, 21, 25, and 30 (4556 observations of 1336 participants; 1999 to 2015) for a total of 11 230 total assessments. Before all interviews, parent and child signed informed consent and assent forms. Assessments were completed via in-person interviews unless participants had moved away from study area, in which case the interview was conducted by phone. Study protocol and consent forms for each assessment were approved by the Duke University Medical Center Institutional Review Board.

All constructs were assessed from participants and a primary caregiver by using the structured Child and Adolescent Psychiatric Assessment.^20,21  This tool focuses on the 3 months preceding the interview to minimize recall bias. Demographic variables included participant sex and race. Race was assessed by parent report as American Indian, Black, and white and included as a potential confounding variable of the association between gun access and adult outcomes. Urban and rural status was based on Census-based designations for the counties and census tracts. For this sample, 10 of the 11 counties (and the census tracts therein) were designated mostly rural in both the 2000 and 2010 Census. The remaining 1 county was home to >40% of the sample and an urban area of ∼280 000.

Up to age 16, participants and primary caregivers were asked whether any household member had a gun, whether the participating child had access to the gun, and whether the participant had his or her own gun (see Supplemental Figure 3 for gun-related assessment questions). In North Carolina, children cannot own a handgun, but they may possess one under adult supervision. In rural areas, such laws are often not strictly enforced. There is no minimum age to possess rifles and shotguns. All participants were categorized on the basis of the highest level of gun access reported by themselves or their parent at any childhood observation: child owning a gun (own gun), having access to a gun in the home (access to gun), having a gun in the home but not having access (gun in home, no access), and not having a gun in the home (no gun in home). No information was collected on whether guns in the home were stored loaded.

Diagnostic and Statistical Manual of Mental Disorders psychiatric status was assessed for emotional disorders (anxiety and mood disorders), behavioral disorders (conduct disorder, oppositional defiant disorder, attention-deficit/hyperactivity disorder), and substance use disorders. Test-retest reliability and validity of Child and Adolescent Psychiatric Assessment diagnoses are similar to other psychiatric interviews.^20,21  The following measures of psychosocial risk were assessed: low family SES (impoverished, low educational attainment, and low occupational prestige), family dysfunction (inadequate parental supervision, domestic violence, high levels of parental conflict, maternal depression, poor marital relationship quality, and high conflict between parent and child), and parent problems (mental illness, substance abuse, or criminality).²² 

All outcomes were assessed by self-report by using the Young Adult Psychiatric Assessment,²³  except for official criminal charges, and included owning and carrying, exposure to gun violence, criminality, and suicidality. Owning and carrying a firearm included owning a gun, carrying a gun, and taking a gun to school or university. Exposure to gun violence includes being present when someone else shot at someone and shooting a gun at somebody (see Supplemental Figure 3 for all gun-related questions). Someone can both be present when someone was shot at (ie, witness) or shoot at someone themselves (eg, as a soldier) without committing a crime. Criminality items included official charges for misdemeanor or felony crimes and were harvested from the Administrative Offices of the Court’s North Carolina database for all participants up to age 25. Also, self-report data were collected on police contact, after having been placed on probation, and after having been incarcerated. Finally, suicidality was assessed through items about passive suicidal ideation, active ideation, or plans for or attempts of suicide as part of the Young Adult Psychiatric Assessment depression module. Adult outcome variables were coded positive if reported at any adult observation.

Each participant was assigned a sampling weight inversely proportional to their probability of selection. All models used the generalized estimating equations option within SAS PROC GENMOD (SAS Institute, Inc, Cary, NC) to derive robust variance (sandwich type) estimates to adjust SDs for the stratified design and multiple observations for each participant. The REPEATED statement was used with data clustered within identifier and by study wave. Associations between variables were tested by using weighted logistic regression. Consistent with common conventions, all percentages provided in the results are weighted, and sample sizes are unweighted. Findings are considered statistically significant at P < .05.

Across assessments, 83% of possible interviews were completed. All 1420 participants were interviewed at least once in childhood (ages 9–16); 1336 (94.0%) were followed-up at least once in adulthood at ages 19, 21, 25, or 30. Missingness of childhood gun access items was rare (∼1%). Childhood gun ownership and access were not associated with attrition in adulthood (P = .35 and P = .57, respectively).

The GSMS includes 1420 participants (49.0% female; n = 630). During childhood from ages 9 to 16, 6674 observations of those 1420 individuals were collected. Fig 1 shows the 3-month prevalence of gun access by age and sex. The overall 3-month prevalence of having a gun in the home was 55.1% (95% confidence interval [CI]: 52.1%–58.7%). Of the children in homes with guns, 63.3% (95% CI: 59.7%–66.9%) had access to a gun, and 25.0% (95% CI: 21.2%–28.8%) owned a gun themselves. The prevalence of having a gun in the home did not change with age (P = .14). However, conditional rates for both gun access and ownership in homes with guns increased with age (P < .001).

Table 1 presents associations among the different levels of gun access and childhood demographic variables, psychosocial risk variables, and psychiatric status. These analyses are based on 6674 observations of 1420 GSMS participants. Each of the gun access groups is labeled with a number that is then used as shorthand for the between-group comparisons in the last 3 columns. Boys were more likely than girls to have access to a gun in the home and to own a gun themselves (P < .001). Gun access differed by race and ethnicity: Black children were much less likely to have a gun in the home or have access to or to own a gun than white children. American Indian children were more likely to have access to a gun in the home compared with white children. Rural children were much more likely to own their own gun or to have access to a gun in the home than urban participants. Gun ownership and access were not associated with low SES but were associated with family dysfunction and parental history of problems. Children with behavioral or substance problems were more likely to have access to a gun or own a gun. Combining gun access and ownership categories supported a similar pattern of associations (Supplemental Table 4). Follow-up analyses at the symptom-level supported concurrent associations of gun access and ownership with symptoms of both conduct and oppositional defiant disorder (Supplemental Table 5).

Table 2 shows associations between domestic gun access in childhood and adult outcomes. All participants were categorized on the basis of their highest level of gun access in childhood. Adult outcomes status was aggregated across 4556 observations of the 1336 GSMS participants assessed in adulthood (ages 19–30). All analyses were adjusted for significant childhood correlates of gun access and ownership identified above (Table 1). The goal was to partial the independent effect of childhood gun access on adult behavior.

All levels of childhood gun access, including just having a gun in the home, were associated with adult outcomes. Having a gun in the home only was associated with adult owning and carrying but not with adult suicidality or crime-related outcomes. Childhood access to guns and owning a gun were both associated with higher levels of adult owning and carrying, exposure to gun violence, criminality, and suicidality. In addition, gun access or ownership in childhood was associated with the dual adult outcome of displaying both suicidal behavior and exposure to violence and criminality. In a follow-up analysis we looked at adult outcomes of gun access or ownership by age 13 (Supplemental Table 6). Early exposure had a similar (but attenuated) pattern of associations with adult outcomes compared with accessing or owning a gun by age 16. Finally, follow-up analyses were conducted for adult outcomes at ages 25 and 30 only (ie, the points in time most distal from childhood, Supplemental Table 7). The pattern of associations was similar to that identified in the main analyses, but some associations, particularly related to gun access but not ownership, were attenuated.

Follow-up analyses were conducted to test whether particular groups of children with gun access were more vulnerable to adult outcomes than others. Interaction terms were used to test moderation between gun access variables and sex (Table 3), urban or rural status (Supplemental Table 8), and history of childhood behavior problems (Supplemental Table 9). Table 3 displays the P values for the moderation term and the individual odds ratios within the respective groups (eg, male or female). The association between childhood gun access and adult gun owning and carrying was moderated by sex only: male individuals with gun access had higher levels of owning and carrying in adulthood as compared with female individuals. Adult exposure to gun violence and criminality were associated with being male and from an urban area and a history of behavior problems. The association between gun access and adult suicidality was strongest in those from urban areas with a history of behavior problems. Despite this increased risk, the majority of men, those from urban areas, and those with a history of behavior problems, even those with childhood gun access, did not display criminality or suicidality in adulthood (see percent columns in Table 3, Supplemental Table 8 and 9).

In this community-representative mixed urban-rural sample, most children lived in homes with guns, and most children in those homes had access to the guns (1 in 3 children overall). Children with in-home access to guns were more likely to be male, be white, live in rural areas, and have a history of behavior problems. Having gun access as a child was associated with higher levels of adult gun owning and carrying, exposure to gun violence, criminality, and suicidality even after accounting for childhood correlates of gun access. Importantly, childhood gun access was associated with a 2 to 4 times higher odds of both adult suicidality and gun violence exposure and criminality, outcomes that are a priority for public health efforts. Children particularly vulnerable to long-term outcomes from childhood gun access were male, were from urban areas, and had a history of behavior problems. Even in these groups, however, most children did not display adult exposure to violence, criminality, or suicidality.

Our study findings concerning prevalence of gun ownership in rural areas¹¹  and of gun access by children were consistent with previous work.⁶  The profile of children with gun access is also similar to what has been seen nationally (higher access for male sex, white race, rural areas, and history of behavioral and substance problems).⁷  Our study was the first conducted to examine adult outcomes of childhood gun access. Childhood gun access was associated with adult gun-related outcomes.²⁴  Importantly, this prospective association persisted after accounting for baseline adjustment for individual risk factors. Thus, childhood gun access meets criteria for being a risk factor for adult outcomes and, importantly, is one that can be mitigated.

What mechanisms might explain the association between childhood gun access and adult outcomes? Early gun access in the home may habituate children to guns and gun use in ways that increase the likelihood of their use for both acceptable aims (eg, protection, recreation, hunting) and deleterious ones (eg, suicide, crime). In this view, guns become a familiar and salient tool for many uses. Next, early gun access may promote socialization to gun culture and other gun users. This shared interest may serve as a basis for social groups that support gun use (again, possibly prosocially) or for risk taking and crime. Such socialization may reinforce cognitive notions about the need for gun ownership to negotiate threat and maintain safety. Finally, access to “lethal means” is a key risk factor for more serious considerations of suicide compared with nonaccess.^25,26  These explanations are consistent with our findings that similar levels of gun access may impose differential risk as a function of individual characteristics.

The GSMS is representative of 11 predominantly rural counties of North Carolina and, thus, is generalizable to children from rural areas in the Southeast United States and Appalachia. American Indians were overrepresented in the communities from which the sample was drawn; African Americans are underrepresented. Our results need to be tested in samples representative of other areas of the United States. Previous findings from this sample on suicidality and violence exposure have, however, been similar to those from other representative samples of youth (eg, Lereya et al, 2015²⁷ ; Jaffee et al, 2005²⁸ ). There may be unmeasured selection effects that both predict families providing children with gun access and the adult outcomes tested here (eg, shared genetic liability). Such confounding may affect the associations observed and the strength of the associations. In addition to these limitations, for this study, we relied on multiple informants for information about gun access but primarily on self-report (with the exceptions of official criminal records) for adult outcomes.

In the United States, private gun ownership remains a widely accepted means of recreation, a livelihood for some, and a sense of perceived assurance of the ability to protect oneself and one’s family. For most gun owners, these aims are met without adverse consequences for themselves, their families, and society. For some, however, discernable individual and social-environmental risk factors, beginning in childhood, may significantly increase the likelihood that they or their children will harm themselves or others with a firearm. The goal of effective gun policy is to balance the individual constitutional rights of citizens with aggregate costs to society. This study is the first to suggest that parental socialization of children to guns can have long-term effects that persist into adulthood and that put some of these same children and others that they come in contact with at risk many years later. This pattern of findings provides a rationale for targeted gun policy around childhood gun exposure and access, including the possibility of legal restriction of firearms access for young adults with a childhood history of crime-related behavioral risk factors for gun violence. There is already evidence that negligence laws, in which the firearm owner is liable for unsafe storage of firearms, are associated with reductions in firearm-related fatalities (homicides, suicides, and accidents) in children.^9,29,30  By identifying particular subgroups of families and children, clearly defined and easily screened, who are vulnerable to early gun exposure, this study also provides further support and rationale for targeted primary-care based screening efforts around gun access and mental health issues in household with children.

CI

confidence interval

GSMS

Great Smoky Mountains Study

SES

socioeconomic status