## Predictors of suicidal ideation and suicide attempts in preadolescent children: A US population-based study

Suicide is a leading cause of death among youth [1], but strikingly little is known regarding predictors of suicide risk in preadolescent children. What research has been done suggests that rates of suicide attempts and deaths in preadolescent children have not decreased in recent years and may even be on the rise [1,2,3,4]. In recognition of this concern, the National Institute of Mental Health has designated reducing childhood suicide as a priority [5] and recently assembled a panel for 2021 focusing specifically on this issue.

Although prior nationally representative studies of suicide have made strides in documenting trends in rates of suicidal ideation, suicidal behavior, and death by suicide among youth, they have largely been primarily conducted with adolescents [6,7,8], examined non-US samples [9, 10], or focused solely on suicide deaths in children [2,3,4, 11]. Suicidal ideation and suicide attempts are clinically relevant in their own right given that experiencing suicidal thoughts and behaviors in childhood is associated concurrently and prospectively with morbidity [9, 12, 13]. In addition, earlier onset of suicidal ideation and behavior may be associated with amplified risk for poor outcomes [13], further justifying the need for nationally representative data on suicide risk in preadolescent children to accurately characterize the scope of this issue.

What non-nationally representative data do exist on factors associated with preadolescent suicide risk suggest that economic disadvantage [14, 15] and female [15] sex may be associated with suicidal ideation and/or behavior, whereas differences based on race are not typically found [16,17,18]. Critically, however, findings have been notably mixed (e.g., O’Leary and colleagues found no significant differences in preadolescent suicidal ideation by sex [16]) and sample sizes small. Prior analysis of Adolescent Brain Cognitive Development (ABCD) study data found sex differences in suicide risk such that there were higher rates of suicidal ideation but not suicide attempts among males compared with females and identified no significant differences in suicidal ideation or attempts by race and ethnicity [19]. Importantly, however, racial and ethnic identities were treated as categories within a single dimension and suicidal ideation was conflated with attempts, preventing conclusions being reached regarding factors uniquely associated with suicide ideation or with attempts among youth with suicidal ideation.

In terms of potential diagnostic predictors, suicidal preadolescent youth were significantly more likely to have a psychiatric disorder and especially multiple psychiatric disorders compared to nonsuicidal preadolescent youth in the Great Smoky Mountains Study [14]. Although depression has been the most consistently studied diagnostic predictor of preadolescent suicide, anxiety disorders, disruptive behavior disorders, and attention-deficit/hyperactivity disorder (ADHD) also may be associated with suicidal ideation and/or behavior in this age group [14, 18]. Again, however, no research has evaluated diagnostic predictors of suicidal ideation and attempts in nationally representative samples of preadolescent youth, and it remains unknown which psychiatric diagnoses are associated with suicidal ideation only or suicide attempts among youth with suicidal ideation.

Data also are needed on rates of treatment utilization among preadolescent children who have experienced suicidal ideation or attempts. Population-based studies with adolescent samples find that among adolescents who have experienced suicidal ideation or who have made a suicide attempt, many do not have contact with mental health providers [20, 21]. It is not yet known, however, what proportion of preadolescent children with a history of suicidal ideation and/or attempts engage with mental health services, nor what factors predict which children receive treatment. Given that delay in receipt of treatment could result in youth not being regularly assessed for imminent suicide risk and learning strategies to reduce this risk, it is vitally important to identify which preadolescent children are not accessing treatment.

The current paper extends prior work, presenting nationally representative data on predictors of suicidal ideation and attempts in preadolescent children in the US from the ABCD Study. Prior studies on suicide risk in the ABCD Study have reported prevalence rates of self-injurious thoughts and behaviors (SITBs), interinformant agreement on reporting of SITBs, and tested whether familial characteristics and dimensions of internalizing and externalizing symptoms predict risk [19, 22]. Here, we build on these studies by evaluating sociodemographic and diagnostic predictors of suicidal ideation and attempts, and by examining predictors of treatment utilization among youth with suicidal ideation and suicide attempts.

## Method

### Sample

The ABCD Study is a longitudinal, multimethod, multi-informant study following children in the US over 10 years to examine mental health trajectories from childhood through adulthood. The present study used the National Data Archive ABCD 2.01 baseline dataset [23], which was collected between 2016 and 2018. The ABCD Study surveyed 11,875 preadolescent children currently residing in the US, all of whom were 9 or 10 years of age at enrollment. Participants were recruited based on age, sex, race, socioeconomic status (SES), and urbanicity for the purpose of matching SES variation in the US. Participants were enrolled at 22 research sites. For each child, one parent or guardian provided informed consent and information about themselves and their child. The design, procedure, and weighting procedures of this study have been reported elsewhere [24,25,26]. All study procedures were approved by The University of California, San Diego institutional review board (IRB), which serves as the central IRB for the study. Children in the weighted sample reflect the larger population of U.S. 9- and 10-year-olds in terms of sociodemographic and geographic factors.

### Measures

#### Suicidal ideation and behavior

Child and parent self-reports of current and past suicidal ideation and behavior were obtained using the computerized version of the Kiddie Schedule for Affective Disorders and Schizophrenia–Present and Lifetime Version for DSM-5 (K-SADS-PL DSM-5) [27]. Items included those assessing passive and active suicidal ideation and suicide attempts. In line with standard clinical practice [28], the child was coded as having lifetime suicidal ideation or having made a lifetime suicide attempt if either the parent or the child endorsed any of the relevant items either currently (i.e., in the past 2 weeks) or in the pastFootnote 1. Variables were then computed to (1) compare children who had only suicidal ideation versus children who reported neither suicidal ideation nor suicide attempts and (2) to compare children who had and had not experienced a suicide attempt among those children with suicidal ideation.

#### Sociodemographic variables

Each child’s parent or guardian completed a demographic questionnaire, which asked about the child’s sex, racial and ethnic identities, family income, parent/guardian education level, and parent/guardian marital status. Youth also reported on their sexual orientation. For the purposes of retaining sufficient power for analyses, categories for race were collapsed into “White,” “Black,” “Multiracial,” and “Other race,” sexual orientation was collapsed into “Gay or bisexual,” “Not gay or bisexual,” and “Did not understand the question,” family income categories were collapsed into five categories ranging from “Less than $25,000” to “$100,000 and greater”Footnote 2, parental education was collapsed into four categories ranging from “Less than high school” to “College graduate,” and parental marital status was collapsed into “Not married” and “married.”Footnote 3

#### DSM-5 mental health disorders and psychiatric treatment utilization

The K-SADS-PL DSM-5 also was used to obtain information on whether each child met lifetime DSM-5 criteria for mental health disorders. Disorders assessed included: major depressive disorder (MDD), separation anxiety disorder, social anxiety disorder, specific phobia, generalized anxiety disorder (GAD), obsessive compulsive disorder (OCD), post-traumatic stress disorder (PTSD), conduct disorder, oppositional defiant disorder (ODD), and attention-deficit hyperactivity disorder (ADHD). Given their extremely low prevalence rates, all eating disorders were collapsed into one category, as were any psychotic disorder. Again, in line with standard clinical practice, a child was coded as having met lifetime criteria for a psychiatric diagnosis if they met criteria based on parent report, child report, or both either currently (i.e., in the past 2 weeks) or in the past.

Parents also reported whether their child had ever received mental health treatment. Specifically, they responded to the question, “Has your child ever received mental health or substance abuse services?” (yes, no, not sureFootnote 4). Although parents also indicted which types of mental health treatment their child received (outpatient, partial hospital, inpatient, outpatient substance abuse, partial hospital for substance abuse, inpatient for substance abuse, psychotherapy, medication management, other), some response categories were too infrequently endorsed for meaningful analyses.

### Statistical analyses

Cross-tabulations were used to estimate the lifetime prevalence of suicidal ideation, suicide attempts, DSM-5 mental health disorders, and treatment utilization. Cross-tabulations also were conducted to produce lifetime prevalence estimates of suicidal ideation and suicide attempts among youth with various sociodemographic characteristics and mental health diagnoses and to estimate prevalence of treatment utilization among youth with suicidal ideation and suicide attempts. A series of bivariate logistic regression analyses were then conducted to test sociodemographic and diagnostic predictors of suicidal ideation, suicide attempts, and treatment utilization among youth with suicidal ideation or suicide attempts. The Benjamini–Hochberg procedure was applied to correct for multiple comparisons in bivariate analyses. Multivariate logistic regression analyses followed in which all sociodemographic factors were included in one model, any diagnosis versus no diagnosis was included in a second model, number of diagnoses (none, single, two, or more) was included in a third model, and all diagnostic factors were included in a fourth model. Sociodemographic factors were covaried in the latter three models. In all cases, the reference group was defined as the category hypothesized to be at lowest risk.

Dependent variables in separate models were suicidal ideation, suicide attempts, and treatment utilization. In the case of analyses with suicidal ideation as the criterion variable, youth with a lifetime history of suicide attempts were excluded to compare “pure” suicidal ideation versus no suicidal ideationFootnote 5. In the case of analyses with suicide attempts as the criterion variable, lifetime history of suicide attempt versus no lifetime history of suicide attempt was compared among youth with suicidal ideation. Treatment utilization was defined as having a lifetime history of any mental health treatment. Results from all regression analyses are presented as odds ratios with 95% confidence intervals. SPSS 27.0 was used to conduct all analyses and weighting procedures were applied to accommodate the complex sampling frame of the survey and to generate nationally representative estimates.

## Results

In the full sample (unweighted N = 11,875), the total lifetime prevalence rates for suicidal ideation was 14.33% (SE = 0.37) and for suicide attempts was 1.26% (SE = 0.12). 3.62% (SE = 0.20) of the sample reported current suicidal ideation and 0.26% (SE = 0.06) reported a current suicide attempt. 13.15% (SE = 0.35) of youth reported a lifetime history of pure suicidal ideation. Among those with a lifetime history of suicidal ideation, 8.70% (SE = 0.82) reported a lifetime history of suicide attempt. Sociodemographic characteristics of the subsamples of youth with lifetime histories of pure suicidal ideation and suicide attempts are found in Table 1.

### Predictors of lifetime “pure” suicidal ideation

Results of the univariate and multivariate logistic regression analyses for sociodemographic predictors of pure suicidal ideation versus no suicidal ideation are presented in Table 2. When considering all sociodemographic variables together, female sex was associated with decreased odds of suicidal ideation (OR = 0.71, 95% CI 0.62–0.81), as was Hispanic ethnicity (OR = 0.73, 95% CI 0.60–0.87), identifying as Black (OR = 0.77, 95% CI 0.62–0.95), and having parents with a high school or GED level education (OR = 0.78, 95% CI 0.62–0.97). Sociodemographic characteristics associated with increased odds of suicidal ideation included identifying as a sexual minority (OR = 3.81, 95% CI 2.49–5.83), multiracial (OR = 1.39, 95% CI 1.13–1.70), having parents who were not married (OR = 1.30, 95% CI 1.10–1.53), and being in the second (OR = 1.29, 95% CI 1.03–1.63) or third (OR = 1.39, 95% CI 1.15–1.69) quintile for family income.

Results of the univariate and multivariate logistic regression analyses for diagnostic predictors of pure suicidal ideation versus no suicidal ideation are presented in Table 3. Having any psychiatric condition was associated with greater odds of suicidal ideation (OR = 2.67, 95% CI 2.32–3.08), with the odds of suicidal ideation increasing when youth experienced two or more disorders (OR = 3.79, 95% CI 3.24–4.43). In the multivariate analysis controlling for sociodemographic variables, the following disorders emerged as significant predictors: MDD (OR = 4.47, 95% CI 3.54–5.64), GAD (OR = 1.85, 95% CI 1.41–2.44), conduct disorder (OR = 1.78, 95% CI 1.28–2.47), ODD (OR = 2.00, 95% CI 1.65–2.41), and ADHD (OR = 1.43, 95% CI 1.20–1.69).

### Predictors of lifetime suicide attempts

Sociodemographic predictors of lifetime history of suicide attempts versus no suicide attempts among youth with suicidal ideation were examined in a series of univariate and multivariate logistic regression analyses (Table 2). In a multivariate model accounting for all sociodemographic factors, the only characteristics that were significant predictors of suicide attempts were sexual minority status and family income. Identifying as gay or bisexual (OR = 2.55, 95% CI 1.02–6.38) and being in the first (OR = 2.86, 95% CI 1.31–6.26) or second (OR = 3.08, 95% CI 1.62–5.86) quintile of family income were associated with increased odds of suicide attempts.

Similarly, univariate and multivariate logistic regressions were used to determine diagnostic predictors of lifetime history of suicide attempts versus no suicide attempts among youth with suicidal ideation (Table 3). At the multivariate level, having a single disorder did not differentiate youth with suicide attempts from those with only suicidal ideation. However, having two or more psychiatric diagnoses significantly increased the odds of suicide attempts (OR = 2.17, 95% CI 1.26–3.72). When accounting for all diagnoses and controlling for sociodemographic variables in a multivariate analysis, only social anxiety (OR = 2.39, 95% CI 1.25–4.58) remained a significant predictor of suicide attempts, whereas separation anxiety was associated with lower odds of suicide attempts (OR = 0.47, 95% CI 0.24–0.90).

### Predictors of psychiatric treatment utilization

Lifetime history of suicidal ideation (OR = 3.48, 95% CI 3.03–3.99) and suicide attempts (OR = 2.30, 95% CI 1.53–3.45) predicted higher odds of receiving mental health treatment. Of youth with pure suicidal ideation, 34.59% (SE = 1.39) received treatment. Among youth with suicidal ideation, 54.82% (SE = 4.91) of youth who had a lifetime history of suicide attempts received treatment. Sociodemographic predictors of treatment utilization in youth with pure suicidal ideation are found in Table 4.Footnote 6 When all factors were considered together in a multivariate analysis, female sex was associated with lower odds of treatment utilization (OR = 0.63, 95% CI 0.48–0.83), as was identifying as Black (OR = 0.57, 95% CI 0.37–0.89) or Hispanic (OR = 0.65, 95% CI 0.43–0.97), and having parents with a high school diploma or GED (OR = 0.45, 95% CI 0.27–0.75). Youth in the lowest quintile for family income had greater odds of receiving treatment (OR = 1.84, 95% CI 1.11–3.05), as did those whose parents were not married (OR = 1.58, 95% CI 1.13–2.21).

Further analyses examined diagnostic predictors of treatment utilization among youth with a history of pure suicidal ideation (Table 5). Multivariate analyses revealed that having one psychiatric diagnosis increased the odds of treatment utilization (OR = 4.89, 95% CI 2.98–8.02), and having two or more diagnoses further increased these odds (OR = 13.84, 95% CI 8.73–21.93). In a multivariate analysis, after controlling for sociodemographic variables, significant predictors included social anxiety (OR = 1.91, 95% CI 1.09–3.34), GAD (OR = 3.08, 95% CI 1.96–4.84), ODD (OR = 2.33, 95% CI 1.65–3.29), and ADHD (OR = 2.11, 95% CI 1.53–2.92).

## Discussion

Data from this nationally representative sample indicate that in preadolescent youth, a lifetime history of suicidal ideation is alarmingly common with suicide attempts more rare, but still of critical concern. The high rates of suicidal ideation and even suicide attempts in this preadolescent sample underscore the need to clarify predictors of suicide risk in children, and in particular, to examine predictors that differentiate children who experience a suicide attempt from those with suicidal ideation. The present study offers unique insight into the predictors of pure suicidal ideation and suicide attempts among preadolescent children in the United States.

Relative to males, females exhibited a lower likelihood of suicidal ideation, but sex was not a significant predictor of suicide attempts. Although rates of suicidal ideation and suicide attempts are consistently higher among female adolescents than male peers [7, 30], this does not appear to be the case in preadolescent children. Prior research with youth in this age group has been mixed regarding sex differences in suicidal ideation with some studies finding females to be more likely to experience suicidal ideation than males [29] and other studies documenting higher rates of suicidal ideation among males than females [13]. In terms of sex differences in suicide attempts, our finding that sex was not associated with having made a suicide attempt is consistent with past research suggesting that sex differences in suicide attempts are not apparent in preadolescent children [29, 31] and may not emerge until adolescence [32]. Our study adds new data from a large, nationally representative sample that emphasizes the need to not overlook the possibility of suicidal ideation in young males. For these males, such early onset of suicidal ideation could be an especially important risk factor for future suicidal behavior, and insofar as early onset of psychopathology is associated with a more severe course, it is possible that they may be highly represented in the group of males who eventually die by suicide. Following preadolescent males with suicidal ideation into later waves of ABCD data will be especially important for better understanding if onset of suicidal ideation at this age, when compared to later onset, is indeed associated with an especially poor long-term prognosis in adolescence and adulthood.

Mounting evidence also demonstrates that sexual minority youth are at elevated risk for self-injurious thoughts and behaviors [33, 34]. Our findings extend this literature, showing that these effects are observable in youth as young as 9- and 10-year-old. In this sample, preadolescents identifying as a sexual minority had almost four times greater odds of experiencing suicidal ideation than those identifying as heterosexual. Among preadolescents with a history of suicidal ideation, those identifying as a sexual minority had over two times greater odds of having made a suicide attempt. Our findings highlight the crucial need for early intervention and prevention efforts aimed at sexual minority youth.

Another group in need of intervention and prevention efforts may be youth whose family income is less than $50,000 per year. In our sample, family incomes of less than$25,000 or $25,000–$49,999 were associated with higher odds of youth having made a suicide attempt. These findings align with previous research evaluating socioeconomic correlates of suicide risk in preadolescent youth. Walsh and colleagues, for example, found that among 11- and 12-year-olds, receiving free or reduced-price school lunch was associated with higher odds of suicidal ideation and suicide attempts among youth with suicidal ideation [15]. Lower-income families face frequent stressors, have fewer resources, and face greater barriers in providing high levels of parental monitoring relative to higher-income families, the latter of which was associated with higher risk of both the suicidal ideation and attempts in previous ABCD Study analyses [19].

In terms of race, the lower odds of suicidal ideation that we observed among Black youth are in line with some previous studies [7]. Recent evidence suggests suicide attempts [35] and deaths by suicide [3] are increasing among Black youth, leading to calls from the NIMH for additional research on this population [5]. Although among youth with suicidal ideation, Black youth did not differ in their odds of having attempted suicide relative to White youth, this does not preclude the possibility that the gap between the rates of suicide attempts and death by suicide among White and Black youth, respectively, may be closing. As the ABCD Study continues, future longitudinal research examining predictors of suicide attempts among Black youth in this sample has promise in informing prevention efforts to mitigate this trend.

It also is important to note that multiracial youth had significantly greater odds of experiencing suicidal ideation, but not suicide attempts, relative to White youth and that Hispanic youth had significantly lower odds of experiencing suicidal ideation, but not suicide attempts, relative to non-Hispanic youth. Prior research with adolescents similarly finds that relative to White youth, multiracial youth are at increased risk for suicidal ideation and suicide attempts [36]. In addition, compared with non-Hispanic individuals, Hispanic individuals are at decreased risk for suicidal ideation and suicide attempts, though this gap appears to be narrowing for youth and may have even reversed in recent years [37]. These findings highlight the importance of future research examining the effects of multiple, and at times intersecting, identities. Racial minority groups experience heightened discrimination, prejudice, and stigmatization, which in turn increases stress and negative mental health outcomes [38]. Intersectionality theory [39] emphasizes that these risks may be further amplified when individuals experience multiple minoritized identities, as in the case of some multiracial youth. Consideration of intersecting racial identities is important for advancing our understanding of suicide risk among youth, as well as informing our perspective for implementing systemic changes to lower risk for youth most in need.

In terms of diagnostic predictors of suicidal ideation and suicide attempts, having at least one psychiatric condition was associated with greater odds of lifetime suicidal ideation and the strongest effects were for youth who had two or more disorders, which predicted both the suicidal ideation and suicide attempts. In fact, youth experiencing comorbidity had nearly four times the odds of a lifetime history of suicidal ideation and over two times the odds of attempting suicide among youth with suicidal ideation. Prior evidence suggests that comorbidity is a robust predictor of suicide risk among both adolescents and adults [7, 40], and the present findings extend this pattern to preadolescent youth. Theories of suicide highlight that greater distress or psychological pain [41] may lead to suicide through the desire to escape this pain [42]. The dose-response relationship observed in our findings and in previous literature may be attributed to the fact that comorbidity is associated with greater levels of distress and impairment [40].

Individual diagnoses were less consistent predictors of suicidal ideation and suicide attempts in this age group. Of particular note is the finding that MDD was associated with substantially increased odds of suicidal ideation but not suicide attempts among those youth with suicidal ideation. That MDD was not significantly associated with suicide attempts is notable because suicidal ideation and behavior are most often thought of as occurring in the context of MDD. The current findings suggest that a more nuanced view of this association may be warranted. That MDD was the strongest diagnostic predictor of suicidal ideation is consistent with this perspective; in fact, the confidence interval for MDD did not overlap with those for any other disorder. In contrast, MDD did not differentiate between the youth with suicidal ideation who had and had not made a suicide attempt. This is in line with a recent meta-analysis that found depression to robustly predict suicidal ideation but only weakly predict suicide attempts among individuals with suicidal ideation [43].

Also of note are the findings that conduct disorder, ODD, and ADHD were associated with suicidal ideation but not suicide attempts in multivariate models. Externalizing disorders often are overlooked when evaluating diagnostic factors associated with suicide risk. In adolescence, however, disruptive behavior disorders, including conduct disorder, ODD, and ADHD were associated with suicidal ideation and attempts in nationally representative data [7]. Findings in this sample of preadolescents suggests a potential pathway in which youth with externalizing psychopathology experience suicidal ideation, which only later in development manifests as suicidal behavior. Specifically, these findings collectively suggest that preadolescent youth with MDD or externalizing disorders may be at particular risk for suicidal ideation, but that psychiatric comorbidity could potentially be even more important in understanding which of these youth may go on to act on their suicidal thoughts. Longitudinal research is needed to test this possibility, however.

We found that 34.59% of preadolescent youth with a lifetime history of suicidal ideation and 54.82% of youth with a lifetime history of suicide attempts received psychiatric treatment. These rates are lower than those documented in studies assessing treatment utilization among adolescents with a history of suicidal ideation or behavior (>80%) [7] and indicate that too few suicidal preadolescent youth receive treatment. One factor may be that preadolescent youth are not disclosing suicidal ideation or behavior to their parents/guardians. This is supported by studies finding of low concordance between youth and parent reports of suicidal ideation and suicide attempts [22, 44, 45].