Use of Tobacco Products and Suicide Attempts Among Elementary School–Aged Children

Key Points Question Is the use of tobacco products associated with an increased risk of self-injurious thoughts and behaviors among children? Findings This cohort study of 8988 preadolescent children enrolled in the Adolescent Brain Cognitive Development study found statistically significant associations between children’s use of tobacco products and suicide attempts. These associations were partially related to increased negative urgency and remained significant after accounting for various demographic, socioeconomic, familial, and clinical risk factors of suicide. Meaning The findings suggest that smoking tobacco products may be a modifiable risk factor that can be addressed in suicide prevention efforts, especially among children.


Introduction
The increasing use of various tobacco products, including e-cigarettes, vapes, and hookahs, among children and adolescents is a critical public health problem worldwide. 16][17] Higher rates of suicide risk are consistently reported among current smokers compared with individuals who quit, 16,[18][19][20][21] underscoring that further investigations are necessary to clarify whether smoking could be a major preventable risk factor associated with suicidal behaviors. 22mpared with adolescents [15][16][17] and adults, 21,23 however, there is a dearth of information available on the association between smoking and suicidal behaviors among preadolescent children.
Many observational studies focused on uncovering risk factors of childhood smoking initiation, including prenatal exposure, 24 psychiatric disorders, 25 family conflict, 24,26 and peer influence. 5veral groups examined brain structures and functional networks for biomarkers of early-onset smoking. 5Others inspected genome-wide data to elucidate the genetic effects shared between smoking and childhood psychopathologic problems. 27Some of these same risk factors (eg, psychiatric disorders, family conflict or dysfunction, peer relationships) have similarly been associated with preadolescent suicidal thoughts and/or behavior. 28However, it remains an open question whether associations between smoking and suicidal behaviors, consistently observed among adolescents and adults who use combustible cigarettes, also manifest among preadolescent children, 29 an increasing number of whom are exposed to emerging tobacco products.
The primary aim of this study was to fill in this major knowledge gap.We analyzed data from the Adolescent Brain Cognitive Neurodevelopment (ABCD) study, a US population-based cohort of 11 878 elementary school-aged children (enrolled at 9-10 years of age).The ABCD data consist of various social, familial, mental, and physical well-being and behavioral measures assessed from baseline and the first 2 years of follow-up (release version 4.0).Using these data, we examined the following 3 questions: (1) Is the use of tobacco products associated with SITBs among children, while accounting for demographic, familial, and socioeconomic confounders?(2) If so, are children's cognition, temperament, and psychopathologic outcomes associated with the link between smoking and SITBs?(3) Are these findings unique to the use of tobacco products or shared with other substance use problems, such as alcohol, cannabis, and prescription drugs use, among children?This research should have an effect on the prioritization of smoking prevention and cessation efforts in health care policies, especially concerning suicide prevention among children and adolescents.

Study Participants
The ABCD study, an ongoing longitudinal study that began in 2017, enrolled a total of 11 868 participants aged 9 and 10 years in 22 US study sites (eTable 1 in Supplement 1).1][32] The ABCD study was approved by a central institutional review board at the University of California, San Diego.All ABCD participants and their caregivers provided written informed assent and consent for human research.This study is the secondary analysis of deidentified ABCD participants and is exempt from informed consent by the Massachusetts General Hospital institutional review board.This report followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

Measures of the Use of Tobacco Products
Children's substance use data were assessed at baseline using the ABCD Youth Substance Use Interview. 29The ABCD Youth Mid-Year Phone Interview for Substance Use was conducted at the 6-month follow-up and then on a yearly basis, asking whether the child has smoked tobacco products at any time in the past 6 months until the interview date.We assembled hair toxicology screening test results from the ABCD Youth Hair Results.Using these data, we generated 3 binary variables for lifetime use of tobacco products corresponding to baseline, 6-month follow-up, and 18-month follow-up.Participants were classified as cases if they ever reported the use of tobacco products in the surveys or had positive test results for cotinine in the hair toxicology tests (eTable 2 in Supplement 1).

Measures of Suicidal Behaviors
Data on SITBs of the participants were obtained from the youth and caregiver reports of the K-SADS-5 (Kiddie Schedule for Affective Disorders and Schizophrenia for the DSM-5).Lifetime nonsuicidal self-injuries (NSSIs), suicide attempts (SAs), and suicidal ideation (SI) were assessed based on the K-SADS-5 suicide survey module (eTable 3 in Supplement 1).Considering the young age of the ABCD study participants, SA cases included participants who reported attempted, interrupted, and/or aborted SA actions with intention to die, while SI cases were participants with passive or active thoughts of suicide but without any SA actions.Nonsuicidal self-injury cases were participants who reported NSSIs without reporting any SA or SI experience (eFigure 1 in Supplement 1).Response rates for the use of tobacco products and SITB outcomes are provided in eTable 4 in Supplement 1.

Additional Measures on Family and Children
To identify and control for potential confounding factors, we examined demographic characteristics, socioeconomic characteristics, family history, and prenatal exposure of ABCD study participants stratified by their use of tobacco products status using the ABCD Parent Demographics Survey, Parent Family History Summary, and Developmental History Questionnaire (eTable 5 in Supplement 1  Prevalence of smoking experience was not statistically different in either sex or across different races and ethnicities.However, children in the case group were exposed to more disadvantaged socioeconomic environments and smoking during pregnancy and were more likely to have parents with depression, alcohol problems, behavioral troubles, and suicide-related incidences compared with children in the control group.Although the prevalence of tobacco product use increased significantly from 1.1% (n = 101) at baseline to 1.7% (n = 151) at 18-month follow-up (χ 2 = 9.66; P = .002),all major characteristics of the group that used tobacco products observed at baseline remained consistent in the follow-up years (eTables 8 and 9 in Supplement 1).

Association of Smoking and SITBs
First, we examined whether the use of tobacco products by children was associated with SITBs.Of 8988 youths, the lifetime prevalence of SITBs at baseline, year 1, and year 2, respectively, were as follows: NSSI, 4.8% (n = 428), 5.7% (n = 508), and 6.2% (n = 560); SI, 12.8% (n = 1151), 16.2% (n = 1454), and 20.6% (n = 1853); and SAs, 1.7% (n = 153), 2.5% (n = 227), and 3.6% (n = 321) (eFigure 1 in Supplement 1).In the basic model, we found statistically significant associations of the use of tobacco products with SAs and SI at all assessed time points (eFigure 3 and eTable 10 in Supplement 1).In the full covariate-adjusted model, the effect size of the use of tobacco products was attenuated, yet it remained independent and statistically significant for both SAs and SI (Figure 1; eTable 11 in Supplement 1).The use of tobacco products assessed at baseline was associated with SAs and SI at baseline, as well as with the outcomes reported in year 1 and year 2.

JAMA Network Open | Psychiatry
Use of Tobacco and Suicide Attempts Among Elementary School-Aged Children ).Similarly, lifetime use of tobacco products assessed at 6-month and 18-month follow-up was associated with the 2 SITB outcomes subsequently measured in year 1 and year 2. In all cases, the estimated effect sizes of the use of tobacco products for SAs were approximately 3 times larger compared with those for SI (Figure 1).Sensitivity analyses confirmed consistent findings across various measures of SITB outcomes (eTables 12-14 in Supplement 1).Despite limited power, we also found prospective associations of the use of tobacco products assessed at baseline and new SA cases reported in follow-up years 1 and 2 (ie, someone who did not report SAs at baseline but did in year 1 and year 2) (eTable 15 in Supplement 1).Conversely, we found no associations of the use of tobacco products with NSSI at all assessed times (Figure 1).

JAMA Network Open | Psychiatry
Use of Tobacco and Suicide Attempts Among Elementary School-Aged Children

Investigation of Children's Cognition and Behavioral Measures
Next, we examined whether the associations between the use of tobacco products and increased risk of SAs and SI are due to children's cognition, temperament, and psychopathology outcomes.Of the 48 measures that we examined in year 2 (eTable 6 in Supplement 1), 13 were significant correlates of the use of tobacco products after multiple testing correction (Figure 2).Children who had reported the use of tobacco products at 18 months were more likely to have conduct disorder and oppositional defiant disorder compared with controls in year 2.These children also exhibited increased levels of rule-breaking behaviors, aggressive behaviors, and social problems relative to controls.In addition, there were notable differences in several impulsivity traits, including a lack of planning, positive urgency, and negative urgency.
However, both bivariate and multivariate regression analyses, which examined the 13 correlates of the use of tobacco products in the same model, found that the association between the use of tobacco products and SAs was independent of all childhood behavioral or psychopathology outcomes (OR, 3.09  In multivariate logistic regression, lifetime use of tobacco products was used as an independent variable, along with other covariates, to measure associations with 3 suicide risk outcome measures: nonsuicidal self-injury (NSSI), suicide attempts (SAs), and suicidal ideation without SAs (SI).Fixed-effects covariates included age, sex, race and ethnicity, parents' highest educational level, parents' marital status, household income, prenatal exposure to smoking, parental history of depression, alcohol problems, behavioral troubles, and suicide.

Comparison With Other Substance Use Data
To evaluate the specificity of the association between the use of tobacco products and increased risk of SAs, we examined the association of SAs and smoking with other types of substance use data.
Along with alcohol, cannabis, and prescription drug use, significant associations between SAs and use of tobacco products remained robust (use of tobacco products and alcohol sipping: OR, 2.70   The use of tobacco products was assessed in 18-month follow-up, while the assessed personality, cognition, and psychopathology outcomes were assessed in year 2. Effect size indicates the regression coefficient β of the exposure (ie, ever use of tobacco products) in multivariate regression, where each child outcome was used as a dependent variable.Additional covariates included age, sex, parent-reported race and ethnicity, parents' highest educational level, household income, parents' marital status, prenatal exposure to tobacco, number of substances used during pregnancy by mother, parental history of depression, alcohol problem, and behavioral troubles.

JAMA Network Open | Psychiatry
Use of Tobacco and Suicide Attempts Among Elementary School-Aged Children

Discussion
Despite a rapid increase in the use of various tobacco products among youths, there have been limited data on its associations with mental health and suicidal behaviors among younger children.
Using 3-year assessment data from a population-based sample of 8988 US preadolescent children, our study provides robust evidence that the use of tobacco products is associated with a 3 to 5 times increased risk of SAs during preadolescence.The heightened risk was detected among children as young as 9 and 10 years and manifested consistently in 2 subsequent years (aged <13 years).This association remained independent of well-established suicide risk factors, including sociodemographic, familial, and children's psychopathology outcomes, as well as other types of substance use (including alcohol, cannabis, and prescription drugs).
Our study advances the field of pediatric suicide research in several important ways.First, to our knowledge, our findings provide the first empirical evidence that an increased risk of SAs, consistently reported for cigarette smokers, [15][16][17] extends to a range of emerging tobacco products and was observable among preadolescent children.In recent years, e-cigarettes, vapes, and flavored nicotine products have gained wide popularity, in part because they are often marketed as safe alternatives to conventional cigarettes. 18,33,348][39][40] Additional work suggests that the association between smoking and SAs may be largest for smoking initiation, 41 which is consistent with the relatively early developmental patterns observed in the present work.
Second, our data extend prior evidence that use of tobacco products by children is a risk factor specifically associated with SAs but not with NSSI or SI.Using data from a community sample of 1458 youths aged 9 to 17 years, Wu et al 17 reported significant associations between cigarette smoking and SAs after controlling for depression, while the previously significant association with SI disappeared.A study by Orri et al 42 reported an association of prolonged exposure to maternal smoking during pregnancy and late childhood with increased risk of children's SAs after accounting for various socioeconomic factors and parents' mental health problems.Likewise, Orri et al 42 did not find associations of maternal smoking with SI.The present study provides additional evidence for this distinction because the association between the use of tobacco products and children's SAs remained robust after accounting for various confounders (including children's own externalizing and internalizing problems), while the association with SI vanished.Our findings thus suggest that information on the use of tobacco products may help distinguish individuals who have SI from those who may act on their thoughts. 20,43,44ird, our data suggest that use of tobacco products was correlated specifically with externalizing problems and temperament issues rather than internalizing problems among this age group.For adolescents and adults, there have been mixed findings about the associations between suicidal behaviors, substance abuse, and psychiatric disorders.Several previous studies noted that smoking among youths with suicidal behaviors is likely to be secondary to affective disorders. 45,46her groups reported that the associations between smoking and SAs were independent of depression and excessive alcohol use among adolescents and adults. 21,47,48Additional work has suggested a particular association with impulsive SAs that is more specific to externalizing symptoms. 49,50MA Network Open | Psychiatry Our findings provide evidence that negative urgency is associated with both smoking exposure and increased risk of SAs among preadolescent children.We found several childhood behavioral outcomes that were significant correlates of the use of tobacco products, many of which, if not all, are established risk factors for substance use problems during adolescence. 51However, negative urgency was the only correlate of the use of tobacco products that remained an independent correlate of SAs after controlling for various suicide risk factors.The robust association between negative urgency and SAs is notable for several reasons.First, this association may be particularly relevant to this age group.This hypothesis is supported by age-related decreases in impulsivity from youth to adulthood 52,53 and evidence that the association between impulsivity and suicide is stronger among youths than adults. 54,55Second, the finding that negative urgency specifically, rather than trait impulsivity more broadly, was a correlate of SAs is consistent with current conceptualizations of suicide risk.Several models of suicide 56,57 posit that proximal risk for this outcome often occurs within circumscribed periods of high arousal.Further work is required to understand the neural mechanisms contributing to the association between smoking, negative urgency, and SAs.

Limitations
Our study should be interpreted with several limitations in mind.1][32] Findings from our study thus may not be generalizable to the entire US population of preadolescent children.
Second, while we have identified significant associations between the use of tobacco products and SAs both cross-sectionally (at baseline) and prospectively (between different assessment events), our observational study design precludes the determination of causal relationships.For our additional thoughts on potential underlying mechanisms between the use of tobacco products and SAs, please refer to the eAppendix in Supplement 1).Third, we used hair toxicology tests to assess tobacco product use, which, while valuable, has limitations in detecting certain nicotine substances and may not capture the full range of tobacco exposure.Environmental factors, particularly exposure to secondhand smoke, might also be associated with the concentrations of substances detected in hair samples, therefore falsely increasing the prevalence of smoking among the study participants.
Fourth, given the young age of the study participants, the prevalence of the use of tobacco products was less than 2%, limiting the statistical power to perform fine-grained secondary analyses, such as investigating the effect of individual types of tobacco products, the role of substance use patterns (eg, quantity, frequency, concentrations, administration routes), and proximal associations between newly developed use of tobacco products and SITBs between various assessment periods.As the ABCD study continues to collect data from adolescents over time, we anticipate that future datasets may offer improved statistical power to enhance our understanding of these critical aspects.Fifth, despite our efforts to control for various confounding factors, including sociodemographic variables, familial influences, and children's psychopathology, the possibility of unmeasured confounders remains, which could influence both tobacco use and suicidal behaviors.

Conclusions
In this study, children reporting tobacco use had an increased risk of SAs but not NSSI or SI.The findings suggest that smoking tobacco products may be a modifiable risk factor for suicide.Taken together, our findings suggest that the presumption that noncombustible tobacco products, including e-cigarettes and vapes, are safer alternatives to conventional cigarettes needs to be more thoroughly investigated, 58 especially in the context of their potential effect on the mental health and suicidal behaviors of children and adolescents.Further research using neurobiology, genomics, and neuroimaging data is warranted to clarify the causal mechanisms underlying the association between the use of tobacco products and the increased risk of SAs through a developmental perspective.

Figure 2 .
Figure 2. Associations of Children's Personality, Cognition, and Psychology Outcomes With the Use of Tobacco Products The category Psychopathology (B) includes binary variables for DSM-5--oriented Adolescent Brain Cognitive Development study Child Behavior Checklist measures.ADHD indicates attention-deficit/ hyperactivity disorder.a False discovery rate (FDR)-corrected P < .001.b FDR-corrected P < .01. c FDR-corrected P < .05.

JAMA Network Open | Psychiatry Use
1, 2022, to September 5, 2023, from the National Institute of Mental Health Data Archive.Comprehensive information regarding sample collection, of Tobacco and Suicide Attempts Among Elementary School-Aged Children JAMA Network Open.2024;7(2):e240376.doi:10.1001/jamanetworkopen.2024.0376(Reprinted) February 26, 2024 2/15 Downloaded from jamanetwork.comby guest on 03/26/2024 ). Parent-reported race and ethnicity information was obtained from the ABCD Parent Statistical analysis was conducted in R, version 4.2.1 (R Project for Statistical Computing).All P values were from 2-sided tests, and results were deemed statistically significant at an FDR-corrected P < .05.

Table 1 .
Major Characteristics of Adolescent Brain Cognitive Development Study Participants Based on Youth-Reported Ever Use of Tobacco Products in the Baseline Abbreviation: GED, General Educational Development.a Cases represent children who acknowledged the use of tobacco products in the past or at present or had positive cotinine results from hair toxicology tests, while controls did not.b Information about the demographic, socioeconomic, e Indicates study participants who selected either American Indian or Alaska Native, Native Hawaiian or Other Pacific, multiple races, or others in the Adolescent Brain Cognitive Development Parent Demographic survey.f Significant after multiple testing correction.
95% CI, 1.70-5.65];P<.001;likelihoodratiotestFDR-corrected P < .001)(Table2).Negative urgency was the only correlate of the use of tobacco products that retained a significant association with SAs independent from other variables (OR, 1.52 [95% CI, 1.31-1.78];P< .001;FDR-correctedP < .001).Smoking data assessed at baseline and 6-month follow-up presented similar findings (eTables 16 and 17 in Supplement 1).In contrast, the association between the use of tobacco products and SI dissipated once childhood behavioral measures were accounted for (eTables 18-20 in Supplement 1).Interaction analyses of smoking with childhood outcome measures did not reveal any significant associations with suicide risk outcomes (eTable 21 in Supplement 1).
Figure 1.Associations of the Use of Tobacco Products With Suicide Risk Outcomes

JAMA Network Open | Psychiatry Use
of Tobacco and Suicide Attempts Among Elementary School-Aged Children Discussion of Hypothesized Neurobiological Pathways and Shared Risk Factors Linking Smoking and Suicidal Behaviors eFigure 1. Number of Study Participants who Endorsed Suicide Attempts (SAs), Ideation (SI), and Non-Suicidal Self-Injuries (NSSI) from the KSADS-5, and Case Group Classifications Used in the Present Study eFigure 2. Flowchart for Study Participant Selection eFigure 3. Association Between the Use of Tobacco Products (UTPs) and Suicide Risk Outcomes Based on the Basic Model eTable 1. List of 22 ABCD Study Enrollment Sites eTable 2. Details of Children's Tobacco Use Outcome Measures eTable 3. Details of KSADS-5 Variables Used for Constructing Lifetime Suicide Risk and Self-Injury Outcome Measures eTable 4. Response Rates for UTP and SITB Measures eTable 5. Data on Sociodemographics, Family History, and Prenatal Exposure to Substance Use eTable 6. Child Behavioral Outcome Measures eTable 7. Details of Children's Substance Use Measures eTable 8. Major Characteristics of ABCD Study Participants Based on Youth-Reported Ever Use of Tobacco Products in the 6-Month Follow-Up eTable 9. Major Characteristics of ABCD Study Participants Based on Youth-Reported Ever Use of Tobacco Products in the 18-Month Follow-Up eTable 10.Association Between UTPs and Suicide Risk Outcomes Based on the Basic Model eTable 11.Association Between UTPs and Suicide Risk Outcomes Based on the All Covariate-Adjusted Model eTable 12. Association Between UTPs and KSADS-5 Youth-Report-Based Suicide Risk Outcomes Based on the All Covariate-Adjusted Model eTable 13.Association Between UTPs and KSADS-5 Parent-Report-Based Suicide Risk Outcomes Based on the All Covariate-Adjusted Model eTable 14.Association Between UTPs and Suicide Risk Outcome Measures Concordant Between KSAS-5 Youthand Parent-Reports Assessed in the All Covariate-Adjusted Model eTable 15.Associations of the Use of Tobacco Products (UTPs) and New SA Cases, While Adjusting for Various Confounding Factors eTable 16.Multivariate Logistic Regression Results for Predicting Suicide Attempts Using UTPs (Both Lifetime Measures Assessed in the Baseline) eTable 17.Multivariate Logistic Regression Results for Predicting Suicide Attempts Using UTPs (UTPs Assessed in 6-Month Follow-Up, SA Assessed in Year 1 Follow-Up) eTable 18. Multivariate Logistic Regression Results for Predicting Suicidal Ideation Using UTPs (UTPs and SI Both Assessed in Baseline) eTable 19.Multivariate Logistic Regression Results for Predicting Suicidal Ideation Using UTPs (UTPs Assessed in 6-Month Follow-Up, SI Assessed in Year 1 Follow-Up) eTable 20.Multivariate Logistic Regression Results for Predicting Suicidal Ideation using UTPs (UTPs Assessed in 18-Month Follow-Up, SI Assessed in Year 2 Follow-Up) eTable 21.Interaction Analysis Results of UTP at Baseline with Children's Cognition, Temperament/Personality, and Psychopathology Measures on SA. eTable 22. Multivariate Logistic Regression Results for Predicting Suicide Attempts (SAs) While Integrating Multiple Types of Substance Use Data in Addition to Smoking Tobacco Products JAMA Network Open.2024;7(2):e240376.doi:10.1001/jamanetworkopen.2024.0376(Reprinted) February 26, 2024 10/15 SUPPLEMENT 1. eAppendix.