Models control for year, individual demographics, county urbanicity, and county youth demographic composition. Data are reported compared with the lowest poverty concentration (0%-4.9% of the county population living below the federal poverty level). Error bars indicate 95% CIs.
eTable. Suicide Methods Among Children 5-19 Years Old, 2007-2016
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.
Hoffmann JA, Farrell CA, Monuteaux MC, Fleegler EW, Lee LK. Association of Pediatric Suicide With County-Level Poverty in the United States, 2007-2016. JAMA Pediatr. 2020;174(3):287–294. doi:10.1001/jamapediatrics.2019.5678
What is the association between pediatric suicide and county-level poverty in the United States?
In this cross-sectional study of 20 982 US youths aged 5 to 19 years who died by suicide from 2007 to 2016, the adjusted pediatric suicide rate increased with increasing county poverty concentration, defined as the percentage of the county population living below the federal poverty level.
The findings suggest that higher county-level poverty concentration is associated with increased suicide rates among persons aged 5 to 19 years, which may have implications for suicide prevention efforts.
Suicide is the second leading cause of death among youths aged 10 to 19 years in the United States, with rates nearly doubling during the past decade. Youths in impoverished communities are at increased risk for negative health outcomes; however, the association between pediatric suicide and poverty is not well understood.
To assess the association between pediatric suicide rates and county-level poverty concentration.
Design, Setting, and Participants
This retrospective, cross-sectional study examined suicides among US youths aged 5 to 19 years from January 1, 2007, to December 31, 2016. Suicides were identified using International Statistical Classification of Diseases, Tenth Revision, Clinical Modification codes from the Centers for Disease Control and Prevention’s Compressed Mortality File. Data analysis was performed from February 1, 2019, to September 10, 2019.
County poverty concentration and the percentage of the population living below the federal poverty level. Counties were divided into 5 poverty concentration categories: 0% to 4.9%, 5.0% to 9.9%, 10.0% to 14.9%, 15.0% to 19.9%, and 20.0% or more of the population living below the federal poverty level.
Main Outcomes and Measures
The study used a multivariable negative binomial regression model to analyze the association between pediatric suicide rates and county poverty concentration, reporting adjusted incidence rate ratios (aIRRs) with 95% CIs. The study controlled for year, demographic characteristics of the children who died (age, sex, and race/ethnicity), county urbanicity, and county demographic features (age, sex, and racial composition). Subgroup analyses were stratified by method.
From 2007 to 2016, a total of 20 982 youths aged 5 to 19 years died by suicide (17 760 [84.6%] were aged 15-19 years, 15 982 [76.2%] male, and 14 387 [68.6%] white non-Hispanic). The annual suicide rate was 3.35 per 100 000 youths aged 5 to 19 years. In the multivariable model, compared with counties with the lowest poverty concentration (0%-4.9%), counties with poverty concentrations of 10% or greater had higher suicide rates in a stepwise manner (10.0%-14.9%: aIRR, 1.25 [95% CI, 1.06-1.47]; 15.0%-19.9%: aIRR, 1.30 [95% CI, 1.10-1.54]; and 20.0% or more: aIRR, 1.37 [95% CI, 1.15-1.64]). When stratified by method, firearm suicides had the strongest association with county poverty concentration (aIRR, 1.87; 95% CI, 1.41-2.49) in counties with 20% or higher poverty concentration compared with counties with 0% to 4.9% poverty concentration.
Conclusions and Relevance
The findings suggest that higher county-level poverty concentration is associated with increased suicide rates among youths aged 5 to 19 years. These findings may guide research into upstream risk factors associated with pediatric suicide to inform suicide prevention efforts.
Suicide is the second leading cause of death among youths aged 10 to 19 years in the United States.1 The Centers for Disease Control and Prevention (CDC) reported that 3013 youths aged 5 to 19 years died by suicide in 2017, equivalent to 4.8 deaths per 100 000 youths. From 2007 to 2017, pediatric suicide rates increased 88%.1 Knowledge of the risk factors for pediatric suicide may guide targeted interventions because effective programs exist to reduce suicidal ideation and attempts.2
The association of poverty with suicide among youths is not well understood. Poverty has been associated with worse youth health outcomes, including higher rates of intentional and unintentional injuries.3,4 Youths experience poverty at the household and neighborhood levels, and each type of poverty may distinctly contribute to youth health outcomes.5 At the household level, receipt of public assistance in Sweden has been associated with a 2-fold higher risk of suicide among adolescents and young adults.6 Neighborhood poverty in Canada has been associated with youth suicidal thoughts and attempts, even after controlling for family socioeconomic status, but the study did not examine suicide deaths.7 Whether socioeconomic deprivation at the community level increases the risk of pediatric suicide has not been well studied.
Suicides have been found to cluster geographically, with mixed associations found between area socioeconomics and suicide rates.8 At the county level in the United States, high-risk areas for suicides in Ohio were associated with higher socioeconomic deprivation.9 In contrast, an analysis10 in Florida found the reverse, with areas at high risk for suicide associated with lower economic deprivation. These studies8-10 were not national samples and did not focus on the pediatric population. The objective of our study was to examine the association of pediatric suicide and county-level poverty concentration in the United States. We hypothesized that there would be higher suicide rates among youths living in counties with higher poverty concentrations.
We conducted a retrospective, cross-sectional study of suicides among US youths aged 5 to 19 years from January 1, 2007, to December 31, 2016. We used mortality data from the CDC’s Compressed Mortality File (CMF), an administrative database composed of death certificate data maintained by the National Center for Health Statistics.11 Causes of death are classified according to the International Statistical Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM), with county-level death counts reported annually.12 Our data use agreement with the National Center for Health Statistics prohibited efforts to determine case identities, which enabled full data set access without data suppression for low counts. From the US Census Bureau, we obtained annual intercensal county population estimates.13,14 We obtained county poverty measures from the US Census Bureau Small Area Income and Poverty Estimates Program.15 We obtained urbanicity measures from the US Department of Agriculture Economic Research Service.16 Data analysis was performed from February 1, 2019, to September 10, 2019. Because the study included only fatalities, it did not meet the definition of human subjects research and was deemed exempt from institutional review board approval by Boston Children's Hospital.
The study outcome was suicide deaths among youths aged 5 to 19 years, as identified within the CMF by ICD-10-CM cause of death codes for intentional self-harm (suicide): U03, X60-X84, and Y87.0. The use of ICD-10-CM codes to identify causes of death has been previously described.17 Suicide methods were identified using the External Cause of Injury (E-code) mortality matrix, which reports mechanism of death and intent (eg, firearm, suicide) based on ICD-10-CM codes.17
Annual county poverty concentration, the percentage of the population living below the federal poverty level, was divided into 5 groups (0%-4.9%, 5.0%-9.9%, 10.0%-14.9%, 15.0%-19.9%, and ≥20%) based on previously published studies.4,18-20 The federal poverty level was $21 027 in annual household income for a family of 4 in 2007 and $24 339 in 2016.21
From the US census, we obtained annual county-level population estimates stratified by race/ethnicity (white non-Hispanic, black non-Hispanic, Hispanic, and other race/ethnicity), age group (5-9, 10-14, and 15-19 years), and sex (male or female). Youths of other race/ethnicity combinations (including non-Hispanic Asian or Pacific Islanders, American Indians, and Alaska Natives) comprised only 9% of the population. In the CMF, 52 youths (0.2%) who died by suicide had unknown Hispanic ethnicity recorded on their death certificates, and these youths were analyzed in the other race/ethnicity category. We determined the annual percentage of youths in each county who were 10 to 14 years of age and 15 to 19 years age; the percentage who were non-Hispanic white, non-Hispanic black, and Hispanic; and the percentage who were male as measures of youth demographic composition by county. Because rural location is a risk factor for pediatric suicide, we included measures of county urbanicity.22,23 Counties were classified into 9 groups using 2013 Rural Urban Continuum Codes, based on population size and adjacency to metropolitan areas.16
We determined the frequencies of suicide deaths among US youths aged 5 to 19 years. We calculated annual suicide rates per 100 000 youths across strata of demographic variables (age, sex, and race/ethnicity) and by the 5 county poverty concentration categories. We determined the relative risk of suicide within each demographic category by calculating adjusted incidence rate ratios (aIRRs) with 95% CIs, using referent categories of age of 5 to 9 years, male sex, and white non-Hispanic race/ethnicity. We assessed for trends in suicide rates over time by estimating a Poisson regression with suicides as the outcome, year as the predictor, and total population as the offset. We determined frequencies of suicide by each method based on the E-code mortality matrix.
To analyze the association between pediatric suicide rates and county poverty concentration, we used a population-averaged generalized estimating equation. We constructed panels to account for clustering of data within counties across years. Each demographic group by race/ethnicity (white non-Hispanic, black non-Hispanic, Hispanic, and other), age group (5-9, 10-14, and 15-19 years), and sex (male or female) within each county was considered a panel, which was then represented by 10 observations, with 1 for each year of the study (2007-2016). County poverty concentration, which could vary each year, was analyzed using the subgroups described above. We estimated a multivariable negative binomial regression model with suicide count as the outcome and county poverty concentration as the independent variable, controlling for year (modeled as a set of dummy variables with 2007 as the referent), county urbanicity, demographic characteristics of the youths who died by suicide (age, sex, and race/ethnicity), and county youth demographic composition. Models included the log of the stratum-specific population estimate as the offset (coefficient constrained to 1) to estimate incidence rates (in which strata are defined by county, year, age group, sex, and race/ethnicity). Thus, each stratum contributed a count of suicide decedents and a corresponding population estimate to the calculation of incidence rates. County poverty concentration was modeled with indicator variables with the lowest poverty concentration (0%-4.9%) as the referent.4,18,19 We calculated aIRRs with 95% CIs for each subsequent poverty concentration. Models used robust SE estimates and an unstructured within-group correlation structure. As a collinearity diagnostic metric, we calculated the variance inflation factor among the independent variables.
For secondary analyses, we decided a priori to select all suicide methods with a total count of 500 or more (ie, a mean of at least 50 incidents per year) during the study period. We did not assess less common methods because these may be underpowered to detect significant differences in suicide risk between poverty concentration categories. To determine the association between pediatric suicide by each method and county poverty concentration, we constructed regression models identical to the one described above, except we used the suicide count for each method as the dependent variable.
We performed exploratory analyses to assess whether the magnitude of the association of pediatric suicide with county poverty concentration was modified by the youth’s sex, age, or race/ethnicity. For these analyses, we added interaction terms between county poverty concentration and each of these demographic factors to the model. For instance, to assess whether the youth’s sex moderated the association between poverty and pediatric suicide, we added an interaction term between poverty concentration and sex. For age, because of convergence issues, a binary age category of 5 to 14 years vs 15 to 19 years was used. All analyses were performed with Stata software, version 13.0 (StataCorp).
From 2007 to 2016, there were 20 982 suicides among youths aged 5 to 19 years (17 760 [84.6%] were 15-19 years of age, 15 982 [76.2%] male, and 14 387 [68.6%] white non-Hispanic) (Table 1). The annual national suicide rate for the study period was 3.35 per 100 000 youths aged 5 to 19 years. The suicide rate for male youths was more than 3 times greater than for female youths (IRR, 3.05; 95% CI, 2.76-3.38). Compared with white non-Hispanic youths, black non-Hispanic youths (IRR, 0.56; 95% CI, 0.48-0.64) and Hispanic youths (IRR, 0.51; 95% CI, 0.45-0.58) had lower suicide rates. The overall suicide rate increased from 2.66 per 100 000 youths in 2007 to 4.12 per 100 000 youths in 2016, with an increase of 4.6% annually (IRR, 1.05; 95% CI, 1.04-1.05). The 3 most common methods of suicide during the study period were suffocation (9631 [45.9%]), firearms (8619 [41.1%]), and poisoning (1282 [6.1%]) (eTable in the Supplement).
The youth demographic composition varied by county poverty concentration, with a lower percentage of white non-Hispanic youth living in counties with higher poverty concentrations (Table 2). Counties with the lowest poverty concentration (0%-4.9%) had a suicide rate of 3.18 per 100 000 youths, and counties with the highest poverty concentration (≥20%) had a suicide rate of 3.35 per 100 000 youths (Table 3).
In the multivariable model, there was a higher suicide rate in the highest poverty concentration counties (≥20%; aIRR, 1.37; 95% CI, 1.15-1.64) compared with the lowest poverty concentration counties (0%-4.9%), and suicide rates increased in a stepwise manner as poverty concentration increased (Figure and Table 4). In this model, male youths had a higher suicide rate than female youths (aIRR, 3.03; 95% CI, 2.92-3.15). Compared with younger youths (5-9 years of age), suicide rates were higher among youths 10 to 14 years of age (aIRR, 56.03; 95% CI, 41.89-74.96) and 15 to 19 years of age (aIRR, 300.05; 95% CI, 224.71-400.64). Compared with white non-Hispanic children, suicide rates were lower for black non-Hispanic children (aIRR, 0.64; 95% CI, 0.60-0.68), Hispanic children (aIRR, 0.61; 95% CI, 0.57-0.64), and children of other races/ethnicities (aIRR, 0.59; 95% CI, 0.55-0.63). Suicide rates were higher in 2016 compared with 2007 (aIRR, 1.53; 95% CI, 1.43-1.64). The suicide rate was higher in the most rural counties compared with the most urban counties (aIRR, 1.66; 95% CI, 1.45-1.91). There was no collinearity between the independent variables.
The association of county poverty concentration with pediatric suicide rates varied by suicide method. Three methods with a total suicide count of 500 or greater during the study period were selected for subanalysis: suffocation, firearms, and poisoning (Table 4). Among firearm suicides, the suicide rate increased in a stepwise manner with increasing poverty concentration compared with the lowest poverty concentration, with the greatest association found in the 20% or greater poverty concentration (aIRR, 1.87; 95% CI, 1.41-2.49). No association was found between suicides by suffocation or poisoning and county poverty concentration.
In our exploratory models, the youth’s sex did not significantly moderate the association of county poverty concentration with pediatric suicide rates. The association between suicide and poverty concentration for youths 5 to 14 years of age (aIRR, 2.02; 95% CI, 1.31-3.13 in the highest compared with the lowest poverty concentration) was greater than the association among children 15 to 19 years of age (aIRR, 1.24; 95% CI, 1.04-1.49; interaction effect: aIRR, 1.63; 95% CI, 1.02-2.59). The association between suicide and poverty concentration for children of other races/ethnicities (aIRR, 2.97; 95% CI, 1.79-4.94 in the highest compared with the lowest poverty concentration) was greater than the association of suicide and poverty concentration among white non-Hispanic children (aIRR, 1.22; 95% CI, 1.00-1.48; interaction effect: aIRR, 2.44; 95% CI, 1.43-4.17). The association between suicide and poverty concentration was not significantly different for black non-Hispanic children or Hispanic children compared with white non-Hispanic children.
In this study, we found that higher suicide rates among US youths aged 5 to 19 years were associated with increasing county poverty concentration in a stepwise manner, after controlling for demographic characteristics and county urbanicity. County poverty concentration was associated with higher pediatric suicide rates by firearms but not by suffocation or poisoning. We found a 57% increase in suicide rates from 2007 to 2016 among youths aged 5 to 19 years. More recent data from 2017 suggest that suicide rates are continuing to increase.1
Particular groups of youths are differentially affected by suicide, including youths aged 15 to 19 years (relative to younger youths), male youths, and white youths.1 Suicides cluster geographically, with a prior study8 demonstrating mixed associations between area socioeconomics and suicide rates. Studies performed at more granular geographic levels of analysis, such as the county or community level, have been more likely to demonstrate an inverse association, with higher suicide rates in lower socioeconomic areas.8 Limited prior studies6,24,25 have examined suicide by socioeconomic status in the pediatric population. Socioeconomic factors previously associated with pediatric suicide include household receipt of public assistance, parental unemployment, and poor parental educational attainment. County-level poverty concentration as an additional associated risk factor may reflect contributions from individual household poverty and neighborhood variables, which we were unable to differentiate in this study.
Poverty is associated with worse health outcomes in children, including infant mortality, child abuse, and injury.3,4,18 Children living in poverty experience higher rates of chronic disease, including more frequent asthma attacks, increased lead levels, and higher rates of substance abuse.26,27 Neighborhood poverty has an association with child health outcomes beyond that of individual family poverty.28 Children born into neighborhood poverty, but not into household poverty, are at higher risk for externalizing mental conditions than children not born into poverty.5 For suicide specifically, a Canadian study7 found neighborhood poverty to be associated with suicidal thoughts and attempts, even after controlling for family socioeconomic status and individual child characteristics.
Poverty could be associated with an increased suicide risk through a variety of potential mechanisms. It may affect early brain development through long-term exposure to toxic levels of stress, with neurobiological stress–mediated pathways ultimately leading to impaired decision-making, behavioral self-regulation, and mood or impulse control.27,29 Multiple childhood adversities are more commonly experienced by children living in poverty, and these adversities have been associated with increased suicide risk in adolescents and young adults.6,30 Children living in poor neighborhoods are exposed to more family turmoil, violence, social isolation, and lack of positive peer influences.27,28,31 Youth living in poor neighborhoods also experience increased hopelessness; more internalizing problems, such as depression and anxiety; and increased externalizing mental conditions.5,32,33 Furthermore, areas of concentrated poverty may lack infrastructure, such as quality schools, sustainable jobs, health care facilities, and mental health resources, that support good health.27 Future research should work to delineate which of these factors contributes most strongly to the increased risk of pediatric suicide in counties with a high poverty concentration.
In our study, we found that the association of county poverty concentration with pediatric suicide varied by method. We stratified by method in the analyses because we hypothesized that the type of suicide method could vary by poverty concentration. In addition, understanding potential differences in suicide methods could be used to inform suicide prevention strategies.34 Of note, we found an association with poverty concentration only for firearm suicides. Firearms are the most lethal means of suicide when attempted.35 Four-fifths of adolescent suicides by firearms occur in the adolescent’s own home, and most adolescents use firearms owned by the parents.36 The presence of a firearm in a household is associated with a significant, but modifiable, risk for suicide among adolescents.37,38 Measures to increase safe firearm storage (such as keeping guns locked and unloaded, with ammunition locked and away from the firearm) are associated with reduced risk of youth unintentional injury or suicide by firearms.39,40 Because areas of high poverty concentration also experience higher rates of unintentional firearm deaths, further study is needed to determine whether children living in these areas have increased access to firearms or less safe firearm storage.18
The increased risk of pediatric suicide in high poverty areas may have implications for suicide prevention efforts. Effective interventions have been delivered in school, community, and health care settings for universal programs and targeted programs for high-risk individuals.2 For instance, the school-based program Saving and Empowering Young Lives in Europe reduced suicide attempts and severe suicidal ideation by approximately 50% at a 12-month follow-up in a multicenter cluster randomized clinical trial.41 The Zero Suicide Approach is an effective model for health care settings that incorporates commitment by health care leadership, practitioner training, attention to screening and assessment, use of a systematic suicide care protocol, and evidence-based treatment strategies.42 We found similar unadjusted rates of pediatric suicide among the different county poverty concentrations; thus, equal allocation of prevention resources regardless of county poverty concentration could be considered. Nevertheless, prevention efforts must also address upstream risk factors for suicide. The independent association between county poverty concentration and pediatric suicide in the multivariable model should prompt research into potential mediators through which poverty might convey an elevated suicide risk. It is unknown whether focused implementation of suicide prevention programs in high poverty areas could have potential for increased yield. On a more societal level, strategies that combat poverty, such as increasing the minimum wage, may be part of a multipronged strategy to consider for reducing youth suicide.43,44
Our results must be considered in the context of the limitations of our study. Because the CMF is an administrative database, there is potential for misclassification of demographics or cause of death. Current estimates of demographic misclassification on death certificates are as high as 40% for American Indian or Alaska Native individuals, 3% for Hispanic individuals, and 3% for Asian or Pacific Islander individuals.45 Attribution of deaths as intentional or unintentional may sometimes be difficult to determine.46 We were unable to analyze data on a more granular geographic level, such as census tracts, which may be more sensitive to economic differences than counties or zip codes.20,26 Counties are heterogeneous and may consist of smaller communities of varying poverty concentration (ie, a city and its suburbs). As a result, the county poverty concentration reflects a combination of its communities, which may underestimate associations of poverty with suicide rates. Because markers of individual household socioeconomic status were unavailable in the databases used, we were unable to determine what proportion of the increased suicide risk was attributable to individual household poverty vs neighborhood-level poverty. We were also unable to explore the mechanisms connecting area poverty and suicide. Because the study was conducted in the United States, our findings may not be generalizable to other regions of the world.
In the United States, increased suicide rates among children were associated with increasing levels of county poverty concentration. This association was most prominent for firearm suicides. As pediatric suicide rates in the United States continue to increase, understanding of the upstream contributors to pediatric suicide, including poverty-related factors, appears to be needed so that suicide prevention efforts can focus on the youths at highest risk.
Accepted for Publication: September 16, 2019.
Corresponding Author: Jennifer A. Hoffmann, MD, Division of Emergency Medicine, Ann & Robert H. Lurie Children’s Hospital of Chicago, 225 E Chicago Ave, PO Box 62, Chicago, IL 60611 (firstname.lastname@example.org).
Published Online: January 27, 2020. doi:10.1001/jamapediatrics.2019.5678
Author Contributions: Dr Hoffmann had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Hoffmann, Farrell, Fleegler, Lee.
Acquisition, analysis, or interpretation of data: Hoffmann, Monuteaux, Fleegler, Lee.
Drafting of the manuscript: Hoffmann, Fleegler, Lee.
Critical revision of the manuscript for important intellectual content: Farrell, Monuteaux, Fleegler.
Statistical analysis: Hoffmann, Monuteaux, Lee.
Administrative, technical, or material support: Fleegler, Lee.
Supervision: Fleegler, Lee.
Conflict of Interest Disclosures: Dr Fleegler reported receiving honoraria for grand round presentations related to firearm injuries, social determinants of health, and sedation and serving as a consultant to Veta Health. Drs Fleegler and Lois Lee received royalties for authoring chapters for UpToDate (publisher, Wolters Kluwer). No other disclosures were reported.