Geomasking was used to protect school confidentiality. Points on map are in similar regions as actual school, but do not represent the actual location of schools.
Surface smoothed using empirical Bayesian kriging in ArcGIS Pro version 4.6.0. Estimated prevalence could only be interpolated prevalence within the boundaries of the farthest Colorado schools, thus the surface ends short of the state line. Different intervals for each measure were created to highlight areas with relative higher and lower prevalence.
Schools meeting this definition had both prevalence of easy handgun access and prevalence of students planning a suicide attempt in the last year in the top quartile. High-risk counties are pink.
eTable 1. Urban-Centric Locale Definitions
eTable 2. Demographic and Geographic Weighted Prevalence of the Sample
Customize your JAMA Network experience by selecting one or more topics from the list below.
Identify all potential conflicts of interest that might be relevant to your comment.
Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.
Err on the side of full disclosure.
If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.
Not all submitted comments are published. Please see our commenting policy for details.
Spark TL, Wright-Kelly E, Ma M, James KA, Reid CE, Brooks-Russell A. Assessment of Rural-Urban and Geospatial Differences in Perceived Handgun Access and Reported Suicidality Among Youth in Colorado. JAMA Netw Open. 2021;4(10):e2127816. doi:10.1001/jamanetworkopen.2021.27816
Are rurality and geography associated with access to handguns and suicidality in Colorado youth?
In this cross-sectional study including 59 556 Colorado high school students, rurality and geography correlated with reported prevalence of easy handgun access and suicidality.
These findings suggest that incorporating geographic information, beyond measures of rurality, could inform where to focus youth firearm safety efforts to prevent suicide.
Suicide is the second leading cause of death in adolescents, with firearms the most common method, especially in rural communities. Identifying where to target lethal means safety interventions could better leverage limited resources.
To understand the associations of rurality, school-level prevalence of easy handgun access, and suicidality measures in Colorado youth, to explore spatial distribution of school-level measures, and to identify communities with high prevalence of both easy handgun access and suicidality.
Design, Setting, and Participants
This cross-sectional study used data from the 2019 Healthy Kids Colorado Survey, an anonymous cross-sectional school-based survey conducted at 256 participating Colorado high schools. Participants included students from schools recruited for statewide population-based estimates and additional schools opting in. Data were analyzed from November 9, 2020, to March 13, 2021.
Urban-centric locale according to a 7-level continuum. Geocoded location of schools was used for spatial analysis.
Main Outcomes and Measures
The main outcomes were weighted prevalence for easy handgun access and 4 measures of mental health and suicidality in the previous year (ie, feeling sad for 2 weeks and considering suicide, planning suicide attempt, or attempting suicide in the past year).
A total of 59 556 students (49.7% [95% CI, 49.3%-50.1%] male and 50.3% [95% CI, 49.9%-50.7%] female; 53.9% [95% CI, 53.5%-54.3%] in 9th and 10th grade; 36.4% [95% CI, 36.0%-36.8%] Hispanic and 50.8% [95% CI, 50.4%-51.2%] non-Hispanic White) from 256 schools participated. Most schools were rural or in small towns (56.8% [95% CI, 50.7%-62.9%]), while more students participated from urban and suburban schools (57.8% [95% CI, 57.6%-58.0%]). Prevalence of perceived easy access to handguns increased with increasing rurality, with 36.2% (95% CI, 35.2%-37.1%) of students in rural (remote) schools reporting easy access, compared with 18.2% (95% CI, 17.3%-19.1%) for city (large) schools. The spatial distribution of easy handgun access and suicidality measures had minimal overlap, but there was correlation at school-level between easy handgun access and considering suicide (ρ = 0.203 [95% CI, 0.0748-0.331]), planning suicide (ρ = 0.300 [95% CI, 0.173-0.427]), and attempting suicide (ρ = 0.218 [95% CI, 0.0869-0.350) in the previous year. The highest quartile for prevalence of both perceived easy access to handguns and planning suicide in the previous year included 21 schools (81.0% [95% CI, 64.0%-97.9%] rural [remote] or rural [distant]).
Conclusions and Relevance
These findings suggest that rural-remote communities in Colorado may benefit most from interventions focused on limiting youth access to handguns when youth are in crisis, with some communities at especially high risk. Spatially referenced data may improve targeting interventions to where they are needed most.
Suicide is the second leading cause of death for youth aged 15 to 19 years in the United States,1 with rates increasing 76% from 2007 to 2017.2 Firearms are the leading method, accounting for 51% of all youth suicide deaths.3 Access to firearms is associated with increased suicide rates.4-10 Although simply owning a firearm is not associated with increased suicidality,11,12 the outcome is more likely lethal when individuals in crisis have access to firearms.13,14 While the best practice is to keep firearms out of the home, households with children opting to own firearms should limit access by storing all firearms locked, unloaded, with ammunition locked and stored separately. However, a 2018 study reported that only 30% of households with children and firearms reported all firearms were stored locked and unloaded.15
Firearm availability and youth firearm suicide risk vary geographically within the US.15-18 Rural youth are at 2-fold increased risk of dying by firearm suicide compared with their urban counterparts.19 Moreover, individuals in rural communities are more than 2-fold more likely than those living in urban communities to own firearms.20 To our knowledge, no surveillance systems track youth firearm access nationally or at smaller scales.21,22
To address this limitation, in 2019, the Healthy Kids Colorado Survey (HKCS), a state-wide biennial cross-sectional survey of middle and high school students, began asking high school students whether they could easily get a handgun if they wanted one. Initial findings indicated that nearly 1 in 5 high school students reported having easy access to handguns.23 However, little is known about youth firearm access across geographic contexts. Identifying and understanding areas where combined high handgun access and suicidality overlap could inform targeted youth-focused interventions to prevent youth firearm access for those at risk of suicide.24
The objectives of this study were to explore the associations of rurality, school-level prevalence of easy handgun access, and suicidality measures in Colorado youth, to examine the spatial distribution of school-level easy handgun access and suicidality measures, and to identify schools where easy handgun access and measures of suicidality were both high. We hypothesized we could identify areas that might benefit most from targeted interventions in collaboration with local school districts and communities.
This cross-sectional study was approved by the Colorado Multiple Institutional Review Board. Participation was voluntary and approved by parents; therefore, participation was considered consent. No identifying information was collected from students. This study is reported following the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cross-sectional studies.
Colorado public high school students received the HKCS survey between August and December of 2019. A portion of survey questions came from the Centers for Disease Control and Prevention’s Youth Risk Behavior Surveillance System, and administration was consistent with Centers for Disease Control and Prevention methods.25 Although the HKCS used a 2-stage stratified cluster sampling design to produce state estimates, any public middle or high school not selected in the state sample could opt-in and have students complete the questionnaire. Data from opt-in schools were included in this study to increase sample size. The anonymous survey was self-administered in classrooms during a regular class period using either a machine-scannable paper booklet or online, as determined by the school.
We used a cross-sectional spatial analysis to explore co-occurrence of different measures across schools and geographies, with no measure treated as exposure or outcome. The measures of interest were perceived handgun access, suicidality, and urban-centric locale.
The 2019 HKCS administration added the question: “If you wanted to get a handgun, how easy would it be for you to get one?” (response options: “Very hard,” “Sort of hard,” “Sort of easy,” and “Very easy”).26 For ease of presentation and analysis, we combined “Sort of easy” and “Very easy” into easy access and “Sort of hard” and “Very hard” into hard access.
Students were asked about their mental health with 4 Youth Risk Behavior Surveillance System questions: “During the past 12 months, did you ever feel so sad or hopeless almost every day for two weeks or more in a row that you stopped doing some usual activities?”; “During the past 12 months, did you ever seriously consider attempting suicide?”; “During the past 12 months did you ever make a plan about how you would attempt suicide?” (all with the response options of yes and no); and “During the past 12 months, how many times did you actually attempt suicide?” with response options of 0 times, 1 time, 2 or 3 times, 4 or 5 times, and 6 or more times, which were collapsed into 1 time or more vs 0 times.
The National Center for Education Statistics Education Demographic and Geographic Estimates program assigned 12 urban-centric locale codes covering 4 types of areas (city, suburb, town, and rural) with 3 subcategories each to every school in the US (eTable 1 in the Supplement).27 Each school’s urban-centric locale assignment was based on actual location, using US Census Bureau’s definition of urbanized areas and urban clusters. We collapsed the 12 categories into 7 owing to small numbers of schools in some categories, keeping schools similar in size and proximity to cities together. This resulted in the following categories: city (large), city (midsize/small), suburb (large/midsize/small) and town (fringe), town (distant/remote), rural (distant), rural (fringe), and rural (remote). School location was obtained from the Colorado Department of Education.28
Descriptive and spatial analyses across urban-centric locale categories were completed to explore statistical and spatial differences in prevalence for key outcome and risk factor variables. Spatial analysis and visualization were used to identify communities at potentially higher risk for youth suicide.
Students were weighted by school enrollment in each participating high school. The weights were constructed to account for sampling design (clustered by classroom), student nonresponse, and discrepancies in grade, sex, and race and ethnicity between the sample and school enrollment. Race and ethnicity were self-reported. In this study, race and ethnicity were used to weight results, describe survey participants, and demonstrate differences across participants in different urban-centric locale. Initial survey analyses were conducted in SAS statistical software version 9.4 (SAS Institute) using SURVEY procedures.
Weighted prevalence and 95% CIs of easy gun access and suicidality (ie, sad for 2 weeks or considering, planning, or attempting suicide in previous year) measures were stratified by urban-centric locale categories. Significant differences between prevalence measures in city (large) schools and schools in other locales were determined using logistic regression (ie, city [large] was the reference group) taking survey design into account. Spearman rank correlation coefficient measured the correlation of easy access to handgun and suicidality measures.29
We created a map to show the spatial distribution of urban-centric locale for each high school. To preserve school confidentiality, geomasking was used to randomly move the location of each school within a maximum distance, so schools appear in their region but not exact location.30
We used global Moran I to statistically test spatial dependence, or whether schools closer together were more likely to have similar values for each variable.31 Moran I is a measure of autocorrelation within 1 variable based on geographic neighbors, in which higher Moran I values indicate that a given value is more likely to be spatially proximate to other values similar to itself. A value of 0 indicates a random patterning of that variable throughout space. For these tests, we used distance-based weight matrices to define neighbors such that each school was assigned at least 1 nearest neighbor, using 42.0 miles as the threshold.32 To calculate significance, we used 1000 Monte-Carlo simulations to estimate P values. For variables with significant global Moran I (P = .05), we calculated local Moran I for each school to identify spatial clusters of high values and low values.33 We sensitivity tested the spatial autocorrelation tests by additionally evaluating Moran I using the minimum distance for all schools to have at least 2 neighbors (44.0 miles). Global and local Moran I were calculated using spdep package in R statistical software version 4.0.2 (R Project for Statistical Computing).
Using empirical Bayesian kriging, we created maps to show the spatial smoothed prevalence of each study variable.34 This technique was used to protect confidentiality of participating schools while visualizing spatial variation.
Schools with a prevalence in the top quartile for both easy handgun access and planning suicide were identified. Counties with at least 1 school meeting these criteria were considered to demonstrate areas with potentially higher risk of youth firearm suicide. Mapping was conducted using ArcGIS Pro version 4.6.0 (Esri).
P values for Moran I were 1-sided, other P values were 2-sided, and statistical significance was set at P = .05 for all tests. Data were analyzed from November 9, 2020 to March 13, 2021.
Colorado has 646 high schools (grades 9-12). Of these, 262 participated in the HKCS in 2019. We used 256 schools in this analysis (5 schools were excluded because they did not have an urban-centric locale designation, 1 school was excluded because the school removed the gun access question). The sample included 59 556 students (49.7% [95% CI, 49.3%-50.1%] male and 50.3% [95% CI, 49.9%-50.7%] female; 53.9% [95% CI, 53.5%-54.3%] in 9th and 10th grade; 36.4% [95% CI, 36.0%-36.8%] Hispanic and 50.8% [95% CI, 50.4%-51.2%] non-Hispanic White) (eTable 2 in the Supplement).
Figure 1 presents a map of participating schools by urban-centric locale. Most schools were rural (distant, fringe, or remote) (56.8% [95% CI, 50.7%-62.9%]) with rural (remote) being the largest category. However, rural (remote) schools only represented 2.9% (95% CI, 2.9%-3.0%) of students across the state with weighted survey results (Table 1). Easy access to handguns increased in a nearly stepwise fashion, from 18.2% (95% CI, 17.3%-19.1%) in city (large) schools to 36.2% (95% CI, 35.2%-37.1%) in rural (remote) high schools, an approximately 2-fold difference (Table 1). Both considering and planning suicide in the previous year were also highest in rural (distant) and rural (remote) schools compared with city (large) schools (considering suicide: 18.6% [95% CI, 17.6%-19.4%] of students in rural [remote] schools vs 16.5% [95% CI, 15.8-17.2%] of students in city [large] schools; planning suicide: 15.6% [95% CI, 14.9%-16.3] of students in rural [remote] schools vs 12.8% [95% CI, 12.1%-13.5%] of students in city [large] schools) (Table 1).
Spearman rank correlation between school-level weighted prevalence of easy gun access and suicidality variables identified significant correlation between school-level prevalence of easy handgun access and considering suicide (ρ = 0.203 [95% CI, 0.0748 to 0.331]; P = .001), planning suicide (ρ = 0.300 [95% CI, 0.173 to 0.427]; P < .001), and attempting suicide (ρ = 0.218 [95% CI, 0.0869 to 0.350]; P < .001) in the previous year. Significant correlation also existed between measures of considering, planning, or attempting suicide (considering and planning: ρ = 0.808 [95% CI, 0.745-0.871]; P < .001; considering and attempting: ρ = 0.717 [95% CI, 0.643-0.792]; P < .001; planning and attempting: ρ = 0.686 [95% CI, 0.600-0.772]; P < .001). We observed no significant correlation between easy handgun access and prevalence of sadness for 2 weeks (ρ = 0.119 [95% CI, −0.122 to 0.250]; P = .06).
All variables, except considering suicide, showed significant spatial autocorrelation (Table 2), suggesting spatial dependence of measures. The highest global Moran I values were observed for weighted prevalence of easy handgun access (I = 0.332; P = .001) and planning suicide had (I = 0.107; P = .001).
Maps visualizing smoothed weighted prevalence for handgun access (Figure 2A) show higher prevalence in both the eastern plains and the mountainous western areas and lower prevalence along the Interstate-25 corridor, where 85% of the Colorado population lives.35 Suicidality measures (excluding considering suicide because it was not spatially dependent) were highest in the south central region of the state, with higher prevalence areas along the western slope of the Rocky Mountains (Figure 2B-D).
Local Moran I results showed a similar spatial pattern to the maps presented in Figure 2. Because of this and confidentiality concerns, we chose not to present separate local Moran I maps but identified 27 schools within significant hot spots with easy handgun access, meaning they, along with neighboring schools, had high prevalence of easy handgun access. Of these, 19 schools (70.4%) were rural (remote) and 8 schools (29.6%) were rural (distant) or town (distant/remote). We identified 78 schools as significant cold spots (37 schools (47.4%) classified as city [large] or city [midsize/small]), meaning that they had low prevalence of easy gun access and were located near other low prevalence schools. There were 9 schools (2 of which were rural), mostly in south central Colorado, that were identified as hot spots for both prevalence of planning suicide and attempting suicide in the last year. Cold spots for these variables were mostly concentrated around the Denver metropolitan area. There were no hot spots for easy handgun access and suicidality, while cold spots for both overlapped in the Denver metropolitan area. Sensitivity analysis indicated no meaningful difference in findings when using a different weight matrix.
Finally, we identified schools with high weighted prevalence of easy handgun access and planning suicide, identifying 21 schools in 19 counties with a prevalence in the top quartile for both measures (Figure 3). Of these, 13 schools (61.9%) were rural (remote), 4 schools (19.0%) were rural (distant), and 4 schools (19.0%) were town (distant/remote).
This cross-sectional study explored school-level prevalence of self-reported ease of access to handguns and suicidality in Colorado youth, analyzing these associations for spatial clustering and by rurality. Consistent with previous findings,36-38 we found that access to firearms increased with increasing rurality. Previous research has reported that firearms are the most lethal method for suicide attempts13,14 and increased access to firearms is associated with higher suicide fatality rates.4-10 Schools in the most remote and rural areas had the highest prevalence of reported easy handgun access. We also identified 19 counties with at least 1 school with highest prevalence of both easy handgun access and planning suicide in the previous year.
We found that rurality was associated with prevalence of easy handgun access but not necessarily with measures of suicidality. Rurality of school was most strongly correlated with students reporting easy handgun access, with more than one-third of students in rural (remote) schools reporting easy access, 2-fold higher than their city (large) counterparts. Because the survey asked youth about their perceived access to handguns, it is unclear whether we captured differences in actual access to handguns. For example, youth could answer yes if there is an unlocked or otherwise accessible handgun in their house, a family member or friend’s house, or they felt they could easily purchase one. The multiple potential reasons for this response would require different community-level interventions. However, since these findings are consistent with previous research into rural-urban differences in firearm ownership,20 this question is likely capturing an important phenomenon.
Some measures of suicidality were highest in rural (remote) and rural (distant) schools, although not in rural (fringe) schools, indicating some, but not all, rural students reported struggles. Research in adults has found increasing suicidality in rural areas, with many possible causes, such as social isolation and low resources (eg, mental health services).17,39-41 The association between rurality and suicidality in youth needs further investigation.
While rurality was strongly correlated with handgun access, looking beyond urban-rural differences and mapping measures at the substate-level provided rich information to inform public health action. To our knowledge, this study is the first to explore the spatial distribution of youth handgun access. Easy handgun access showed more spatial clustering than did suicidality measures, with hot spots of high handgun access around the periphery of the state, predominantly in rural areas. Combining empirical Bayesian kriging with local Moran I allowed us to visualize and identify areas with especially high prevalence of reported easy handgun access.
We did not observe consistent spatial overlap in easy handgun access and suicidality hot spots statewide. However, we identified counties with at least 1 school with both higher prevalence of self-reported easy handgun access and suicidality. Previous research has reported that when a person is suicidal, having easy access to a firearm can be lethal.11,24,42 Therefore, our study contributes to the identification of schools and communities with potential for higher ease of access to firearms when youth are at risk of suicide.
The social ecological model promotes looking beyond individual-level interventions to the community or societal level to achieve population-level impact.43 Using school-level prevalence allowed for identification of communities potentially at highest risk of firearm suicides. These findings draw attention to schools or communities that might benefit most from population-based suicide interventions. Selecting and targeting localities at highest risk could have a bigger impact on health outcomes,44 allowing for directed and tailored firearm suicide interventions to improve the effectiveness and use of limited resources in adolescent suicide prevention efforts.
While understanding rural-urban differences is important for geographically appropriate approaches to prevention, not all rural areas have the same risk. This study demonstrated that using school-level data could better specify where risk of firearm access and suicide were higher.
This study has several limitations. First, data were self-reported by high school students, although research indicates that overall, high school students accurately respond to confidential surveys.25 Second, the survey question specifically asks about access to handguns. In rural counties, long guns are used in half of adolescent suicides, although other studies have reported that handgun use in suicides is increasingly common.37,45 The impact of the wording of the question on findings is not entirely clear; the 2015 National Firearms Survey found that gun owners had a mean of 4.8 guns20 and 75% of gun owners with children in the home owned at least 1 handgun.15 Therefore, it is possible that the results would not differ drastically if the survey question asked about easy access to any firearm type. Additionally, not all Colorado high schools participated in HKCS. Although we identified schools and communities that would likely benefit most from intervention, there could be other schools and regions that also have high measures of handgun access and suicidality that were not surveyed.
Comparison of our findings with other studies is difficult because of both lack of national surveillance data on firearm access and suicide attempts and low base rates of youth suicide mortality, especially at smaller geographies. Therefore, this study provides an important contribution to understanding youth firearm access and suicidality using school-level data to indicate possible school and community risk of youth firearm suicide. Future studies should track firearm access and suicidality measures over time and explore how contextual information may help explain trends. Additionally, measures of firearm access regarding handguns and long guns should be included in other national surveillance systems (eg, Youth Risk Behavioral Surveillance System and Behavioral Risk Factor Surveillance System) to explore patterns and identify higher-risk areas in other states.
This cross-sectional study identified schools and communities with high prevalence of suicidality and handgun access. In Colorado, rural and remote communities with high prevalence of easy handgun access and suicidality may benefit most from reducing access to lethal means when a young person is in crisis. Ongoing investment in interventions across the socioecological model to temporarily reduce youth access to firearms when youth are in crisis is a necessary next step for preventing youth suicides. National surveillance systems should consider including questions about firearm ownership and access to improve research and public health action to prevent firearm-related mortality.
Accepted for Publication: July 30, 2021.
Published: October 8, 2021. doi:10.1001/jamanetworkopen.2021.27816
Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2021 Spark TL et al. JAMA Network Open.
Corresponding Author: Talia L. Spark, PhD, MS, Rocky Mountain Mental Illness Research, Education, and Clinical Care, Department of Veteran Affairs, 1700 N Wheeling St, G3-154, Aurora, CO 80045 (firstname.lastname@example.org).
Author Contributions: Drs Brooks-Russell and Spark had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Spark, Wright-Kelly.
Acquisition, analysis, or interpretation of data: Spark, Ma, James, Reid, Brooks-Russell.
Drafting of the manuscript: Spark, Wright-Kelly, Ma.
Critical revision of the manuscript for important intellectual content: Spark, James, Reid, Brooks-Russell.
Statistical analysis: Spark, Ma, James, Reid.
Obtained funding: Brooks-Russell.
Administrative, technical, or material support: James, Brooks-Russell.
Conflict of Interest Disclosures: None reported.
Funding/Support: Funding for data collection was supported by a contract with the Colorado Department of Public Health and Environment (Dr Brooks-Russell).
Role of the Funder/Sponsor: The Colorado Department of Public Health and Environment had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.