Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Jun 30.
Published before final editing as: Arch Suicide Res. 2026 Mar 27:1–18. doi: 10.1080/13811118.2026.2648604

Growth trajectories of suicide ideation and attempt in a population-based cohort from the United States

Séverine Lannoy 1, Peter B Barr 2, Nathan Gillespie 1, Mallory Stephenson 1, Alexis C Edwards 1
PMCID: PMC13312499  NIHMSID: NIHMS2184429  PMID: 41889303

Abstract

Objective:

Suicidal thoughts and behaviors (STB) constitute high public health concerns. Understanding the longitudinal trajectories of STB from adolescence to adulthood as a function of sex, ancestry, and genetic liability would improve our knowledge and the design of population-based prevention/intervention.

Methods:

We used data from the National Longitudinal Study of Adolescent to Adult Health (Add Health) and included participants of European (EA, N=4,905) and African (AA, N=1,654) ancestry. We evaluated the growth trajectories of suicide ideation and suicide attempt from age 12 to 40 separately in participants of EA and AA and assessed the roles of sex and genetic liability, indexed using polygenic scores (PGS).

Results:

Quadratic growth models including an age-by-sex interaction fit the data best in both ancestral groups. Results indicated an overall decrease of STB from adolescence to adulthood, a stabilization between ages 25-35, and a tendency to increase after age 35. Sex differences were evidenced by higher baseline levels of STB and sharper decreases across ages in females. Including PGS improved the model fit and was related to baseline levels of STB.

Conclusions:

Adolescence constitutes a high-risk period for the development of STB, particularly in females and those with high genetic liability. Though we observed a stabilization of STB in adulthood, another risk period may arise after age 35.

Keywords: Growth modeling, suicide ideation, suicide attempt, sex, genetic

Introduction

Suicidal thoughts and behaviors (STB) are significant public health burdens. In the United States (US) in 2024, 14.3 million adults seriously thought about suicide, and 2.2 million attempted suicide. Suicide is estimated as the eleventh leading cause of death; in 2024, almost 49,000 people died by suicide (Substance Abuse and Mental Health Services Administration, 2025). Evidence suggests that the rate of suicide increased by 37% between 2000-2018. Though we observed a slight decrease in 2020, a new increase in suicide rates was reported in 2022 (Centers for Disease Control and Prevention, 2025), placing the study of STB among the top mental health research priorities in the US (Prabhakar & Rice, 2023).

Research indicates substantial variability in STB risk across ages (Substance Abuse and Mental Health Services Administration, 2025). Recent studies indicate that suicide ideation and self-harm behaviors develop before age 12 (Geoffroy et al., 2022; Liu et al., 2022). The highest rates of STB are currently reported in adolescence and young adulthood (Ivey-Stephenson et al., 2020), with suicide being the second leading cause of death in that age group (Curtin, 2020; Hawton et al., 2012). Adolescence is a risk period for the onset of psychopathology and risky behaviors; according to developmental neuroscience theories, this can be explained by ongoing brain maturation that does not enable proper emotional regulation and triggers risk-taking (Casey et al., 2008; Steinberg et al., 2008). These theoretical perspectives also suggest changes in risk across developmental periods. Accordingly, many studies have been interested in understanding the changes and correlates of STB during adolescence and through the transition to adulthood (Erausquin et al., 2019; Geoffroy et al., 2021; Shi et al., 2022). Although our knowledge of STB has greatly improved with the use of more sophisticated statistical models (Ram & Grimm, 2009), we still lack fundamental information about how STB emerge and evolve from adolescence to adulthood.

Previous studies mainly illustrate the heterogeneity of STB trajectories (Czyz & King, 2015; Geoffroy et al., 2021; Lopez et al., 2023; Nakar et al., 2016; Nkansah-Amankra, 2013; Prinstein et al., 2008; Shi et al., 2022; Wu et al., 2022; Xiao & Lindsey, 2021; Zhu et al., 2019) – i.e., they captured subgroups characterized by different STB trajectories. For example, a large study conducted within a population-based sample in the US (National Longitudinal Study of Adolescent to Adult Health [Add Health]) indicates that suicide ideation from ages 11 to 31 was characterized by three distinct trajectories: individuals with sustainably high ideation, individuals with sustainably low ideation, and those presenting suicide ideation during adolescence only (Erausquin et al., 2019). Suicide attempt in this sample was best represented by two groups showing a declining high attempt risk and an overall low attempt risk (Erausquin et al., 2019). Findings indicate that these subgroups were influenced by sex, race/ethnicity, as well as social and familial environments (Erausquin et al., 2019; Xiao et al., 2021), replicating prior evidence of sociodemographic differences in STB prevalence. A large body of literature indeed shows higher risks of non-fatal STB in females compared to males, as well as in individuals who identify as American Indian or White compared to Black (Bornheimer et al., 2022; Carter et al., 2022; Ivey-Stephenson et al., 2022; Wang et al., 2016; Xiao & Lindsey, 2021). However, to our knowledge, no studies have leveraged genetically inferred ancestry, which offers a complementary perspective by enabling ancestry-specific analyses of growth trajectories and the incorporation of polygenic risk scores into longitudinal models. Genetically inferred ancestry—derived empirically from genetic similarity to reference panels—is correlated with, but conceptually distinct from, categories of race and ethnicity (Pereira et al., 2024; Tang et al., 2005). This approach facilitates a focus on biological patterns without endorsing discrete racial groups.

Although previous studies have provided important insights into the characteristics of at-risk groups based on their developmental trajectories, complementary work is needed to delineate average trajectories within the broader population. Such efforts can help identify universally high-risk developmental periods and clarify predictors of change that may inform population-level public health strategies. Beyond sociodemographic factors, emerging evidence also implicates genetic influences in the etiology of STB, with studies suggesting both common and distinct genetic contributions across different STB phenotypes (Ashley-Koch et al., 2023; A. R. Docherty et al., 2023; Kimbrel et al., 2022; Mullins et al., 2022). Previous studies indicate the role of genetics in the growth trajectory of risky behaviors (Barr et al., 2022; Edwards et al., 2017), but information regarding the effect of genetics across the course of STB is scarce. Twin studies indicate heritability estimates for suicidal behaviors between 30-50% (Edwards et al., 2021; Voracek & Loibl, 2007), with higher heritability among young people relative to adults aged 25 or older (Edwards et al., 2021). Though this suggests potentially distinct effects of genetic factors across ages, the specific role of genetic liability in the trajectories of STB remains to be investigated. This investigation could be pursued in the context of ancestry-specific analyses of growth trajectories.

In this study, we estimated trajectories of suicide ideation and attempt from adolescence to adulthood in Add Health, a population-based cohort. We capitalized on the most recent data release to include participants from 12 to 42 years old. Compared to previous studies using the same sample, this allowed us to expand the follow-up to mid-adulthood. This study was designed to address three major knowledge gaps: (i) What are the overall trajectories of suicide ideation and attempt (STB) from adolescence to adulthood? To answer this, we used linear mixed-effects models to evaluate population-average changes over time. This method is particularly useful for modeling relatively uncommon phenotypes such as suicide attempt and allows the incorporation of growth predictors. We conducted ancestry-specific analyses in participants of European and African ancestry to help determine the specific periods related to high versus low risk in each subgroup and further test the role of genetic liability; (ii) Are STB trajectories similar across sexes? Although sex has been examined in most studies, to our knowledge, few reports have formally tested the moderation between age and sex to guide the modeling of sex-specific trajectories of STB (Madhavan et al., 2021). This approach allows us to detect sex differences in the course of STB – that is, whether the trajectories of males and females are completely different or whether there are some overlaps in the low- or high-risk periods; and (iii) To what extent does genetic liability influence the trajectories of STB? To answer this question, we used measures of aggregate genetic risk (polygenic scores; PGS) computed based on genome-wide association studies (GWAS). In particular, we aimed to test the interaction between age and PGS to explore whether genetic liability is related to baseline levels of STB risk and/or their changes over time. We also aimed to investigate potential sex differences in the effect of genetic risk. Providing information about the overall change in STB according to ancestry, sex, and genetic liability using longitudinal data can lead to important insights for the design of targeted prevention.

Materials & Methods

Participants

Data come from Add Health, a nationally-representative, longitudinal study of US adolescents beginning in 1994 (for more details, see K. Harris, 2013; K. M. Harris et al., 2009; Kathleen Mullan Harris & Udry, 2022). Add Health is a panel study including five waves of data collection starting from adolescence (ages 11-18) to adulthood (ages 32-42). Participants were selected from a stratified sample of 132 schools resulting in an initial, nationally representative, sample of 90,118 students in grades 7-12. Of the original sample, 20,745 were selected for additional in-home interviews. Of those who completed the Wave I interview (1994-1995), 14,738 completed Wave II (1996); 15,197 completed Wave III (2001-2002); 15,701 completed Wave IV (2007-2008), and 12,300 completed Wave V (2016–2018). In addition, 15,159 individuals provided a DNA sample (Wave IV), and were genotyped using the Illumina Omni1 and Omni2.5 arrays. After quality control, genotypic data are available for 9,974 individuals (Braudt & Harris, 2018). Add Health participants provided written informed consent for participation in all aspects of Add Health in accordance with the University of North Carolina School of Public Health Institutional Review Board guidelines that are based on the Code of Federal Regulations on the Protection of Human Subjects 45CFR46 (https://www.hhs.gov/ohrp/humansubjects/guidance/45cfr46.html). The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008.

For the current analyses, we selected unrelated individuals of European (N = 4,905) and African (N = 1,654) ancestry (Braudt & Harris, 2018) from the full database. Attrition is accounted for in sensitivity analyses that include the survey weights provided in Add Health. Ancestry was assigned based on genetic similarity to reference panels by Add health researchers (Highland et al., 2018). This study includes language related to both race/ethnicity, which reflects socially constructed categories, and genetic similarity (“ancestry”), which uses empirical assignment based on available reference panels, because both are relevant for the current outcomes. Here, we follow best practices for current approaches to handling genetic data from diverse populations (Peterson et al., 2019) in order to limit the possibility of false positives due to population stratification (presence of a systematic difference in allele frequencies) and variation in linkage disequilibrium (LD, the correlation of nearby variants in the genome). The mention of both concepts is in no way endorsing the notion that these reflect discrete biological categories. We also note that the terminology around genetic ancestry/similarity is evolving (Committee on the Use of Race, 2023); the current use of ancestry is in line with Add Health descriptions.

Measures

We focused on two primary outcomes: suicide ideation and suicide attempt. Survey items are available from Wave I to V and ask (1) “During the past 12 months, did you ever seriously think about committing suicide?”, and (2) “During the past 12 months, how many times did you actually attempt suicide?” Both items recorded a frequency and were recoded into binary (0: no suicide ideation, 1: suicide ideation) and ordinal (0: no suicide attempt, 1: one suicide attempt, 2: more than one suicide attempt) variables to avoid skewness in the distribution. In Waves I-III, suicide attempt was conditioned on suicide ideation, but in Waves IV and V, all participants were asked about suicide attempt.

We evaluated the possible role of aggregate genetic liability by using polygenic scores (PGS) for suicide ideation (Ashley-Koch et al., 2023) and suicidal behavior (A. R. Docherty et al., 2023). Though these phenotypes are highly genetically correlated (Ashley-Koch et al., 2023), the correlation was significantly lower than 1 (rG = 0.87). This suggests that the etiology of suicide ideation versus behavior does not completely overlap; therefore, these analyses relied on both PGS. Each PGS is computed by weighting the number of “risk” alleles an individual carries at each locus by effect sizes from GWAS summary statistics, and aggerating these across the genome to derive individual-level scores reflecting one’s genetic risk for suicide ideation and suicide attempt. We used PRS-CSx, a Bayesian polygenic method (Ruan et al., 2022), and PLINK 2.0 (Chang et al., 2015) to compute PGS. PRS-CSx allows the joint modeling of the GWAS summary statistics from EA and AA and combines genetic effects across populations using a shared continuous shrinkage prior on the effect sizes of single nucleotide polymorphisms (SNPs; i.e., the genetic variants). We used 1000 Genomes European and African reference panels (Genomes Project et al., 2015) to leverage linkage disequilibrium diversity across samples. Population-specific posterior effect sizes were combined using an inverse-variance-weighted meta-analysis within the Add Health sample (Ge et al., 2022). We used the weighted summary statistics to derive a PGS for each individual in our sample using PLINK 2.0. Each individual has one PGS for suicide ideation and another for attempt. To account for population stratification, we regressed each PGS on the first 10 ancestral principal components (PC) and computed the standardized residuals from the regression model to correct the PGS scores. We then standardized the PGS, so the scores have a mean of 0 and standard deviation of 1; the effect of PGS in subsequent models represents the change in STB per one standard deviation increase in genetic liability.

Lastly, we included age and self-reported sex as covariates. We modeled changes in STB as a function of age. After excluding relatives, we did not have participants younger than 12 years old at Wave I and few people older than 40 at Wave V (N=185). We thus recorded all observations from 12 to 40.

Statistical analyses

To observe the changes in STB over time, we fit a series of linear mixed-effects models separately for suicide ideation and attempt in our two populations (EA and AA) using the lme4 package in R (R Core Development Team, 2021). We followed guidelines on longitudinal data analysis and first designed an unconditional model to quantify STB variation and then tested changes in model fit by adding linear and quadratic functions (Singer & Willett, 2003). We tested 5 nested models: (1) an unconditional model quantifying outcome variation across people without regard to time to estimate the intraclass correlation coefficient (ICC). The ICC is a descriptive statistic (from 0-1) measuring the proportion of the total outcome that lies between people, therefore indicating the strength of individual differences (Singer & Willett, 2003) - values closer to 0 suggest that most variance is within individuals (e.g., fluctuations across ages), while a value closer to 1 indicates that variance is predominantly between individuals (e.g., stable differences among subjects); (2) a linear growth model to quantify outcome variation across both people and time. In this model, the random intercept was represented by participants and we included a fixed linear term for age; (3) a linear growth model including a sex by age interaction, (4) a quadratic growth model, including a random intercept represented by participants and a quadratic term for age to allow for non-linear change; and (5) a quadratic growth model including a sex by age interaction. We chose the best-fitting model according to the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and log-likelihood. We also evaluated significant differences between models using change in log-likelihood. Assuming good model fit with linear or quadratic functions, we did not test alternative specifications. Though adding a cubic function might help provide an accurate representation of the data, the number of time points and current sample sizes would limit its interpretability and increase the risk of overfitting (Wood et al., 2015). Similarly, we considered adding random slopes to the growth models, but in most cases, the model did not converge. To ensure consistency and comparability of estimates across all models and populations, all analyses fit a random intercept model.

After identifying the best-fitting model from above, we evaluated the effect of PGS. Because the literature is scarce on how genetic liability for STB might shift across developmental periods, we fitted and compared 5 additional models to evaluate the role of aggregate genetic liability: (1) adding PGS as a fixed effect, which tests whether PGS influences the baseline level of STB, (2) adding a PGS-by-sex interaction, testing differential effects of PGS across sexes, (3) adding a PGS-by-age interaction, evaluating shifts in the effect of PGS across ages, (4) adding PGS-by-sex and PGS-by-age interactions, and (5) adding a three-way interaction, i.e., PGS-by-sex-by-age, which tests differential effects of PGS across ages and sexes. In the models focused on suicide ideation, we used PGS for ideation, and in the models with suicide attempt as outcome, we used PGS for suicidal behavior.

As we relied on a large sample, we used linear-mixed models with binary (suicide ideation) and ordinal (suicide attempt) outcomes, as linear models are more flexible and provide higher computational efficiency to run and compare the different models. Mixed-effects models can accommodate missing longitudinal data (i.e., by deleting only the missing wave but not all the data from a participant) but do not fully account for missingness (<26% of missing data for SI across waves and < 27% for SA in Waves IV and V). To assess the reliability of our results, we conducted two sensitivity analyses of the best-fitting growth models: (i) using sampling weights to correct for sampling design, and (ii) using logistic-mixed models. Results did not differ substantially (Supplement).

Results

Descriptive analyses

The EA and AA samples were 53% and 56% female, respectively. Table 1 reports the prevalence of STB across waves, ancestry, and sex. The prevalence indicates higher rates of suicide ideation and attempt in participants of EA than AA and an overall decrease from adolescence to adulthood in both groups.

Table 1.

Prevalence of past 12-month Suicide Ideation and Suicide Attempt at Waves I-V in Females and Males of European and African Ancestries (EA and AA).

Outcomes Participants of
European ancestry		 Participants of
African ancestry
TOTAL FEMALES MALES TOTAL FEMALES MALES
Suicide Ideation (yes / no)
  Wave I 700 / 4172 437 / 2163 263 / 2009 191 / 1448 135 / 781 56 / 667
    Prevalence (%) 16.8% 20.2% 13.1% 13.2% 17.3% 8.4%
  Wave II 442 / 3187 280 / 1650 162 / 1537 99 / 1059 69 / 582 30 / 477
    Prevalence (%) 13.9% 16.9% 10.5% 9.3% 11.9% 2.10%
  Wave III 288 / 3763 159 / 2066 129 / 1697 61 / 1253 36 / 720 25 / 533
    Prevalence (%) 0.08% 0.08% 0.08% 4.9% 5% 4.7%
  Wave IV 375 / 4514 199 / 2412 176 / 2102 111 / 1515 78 / 840 33 / 675
    Prevalence (%) 0.08% 0.08% 0.08% 7.3% 9.3% 4.9%
  Wave V 253 / 3373 139 / 1911 114 / 1462 80 / 936 53 / 595 27 / 341
    Prevalence (%) 0.07% 0.07% 0.08% 8.5% 8.9% 7.9%
Suicide Attempt (yes / no)
  Wave I 186 / 512 136 / 300 50 /212 63 / 127 51 / 83 12 / 44
    % within those with ideation 36.3% 45.3% 23.6% 49.6% 61.4% 27.2%
  Wave II 126 / 316 100 / 180 26 / 136 36 / 63 26 / 43 10 / 20
    % within those with ideation 55.8% 55.5% 19.11% 57.1% 60.4% 50%
  Wave III 61 / 226 37 / 122 24 / 104 19 / 42 11 / 25 8 / 17
    % within those with ideation 26.9% 30.3% 23.1% 45.2% 44% 47.1%
  Wave IV 77 / 4814 39 / 2573 38 / 2241 24 / 1604 48 / 904 9 / 700
    Prevalence (%)I 1.6% 1.5% 1.7% 1.5% 5.3% 1.3%
  Wave V 42 / 3577 30 / 2016 12 / 1561 21 / 991 13 / 633 8 / 358
    Prevalence (%)I 0.01% 1.5% 0.77% 2.1% 2.1% 2.2%
Open in a new tab
I

Suicide attempt from Waves I-III in Add Health has only been asked to participants who endorsed suicide ideation. In Waves IV and V, the item related to suicide attempt was asked to everybody in the sample (supporting the evidence that suicide ideation does not always precede suicide attempt), this is why the prevalence differs substantially. We chose to not recode prevalence at Waves I-III, as we cannot assume that because participants were not asked about suicide, they did not report suicide attempt. This, however, does not impact the statistical analyses. Table 1 is for descriptive purposes.

Trajectories of suicide ideation

EA group.

From the unconditional model, we found an ICC of 0.161, suggesting within-individual variance in suicide ideation over time. Table 2.A. indicates the fit statistics for the 5 models evaluating the growth trajectories of suicide ideation. The quadratic growth model including a sex-by-age interaction fit the data best (AIC= 7599.6, BIC = 7663.3). Results from this model (Table S1; Figure 1, part A) showed sex-specific effects. In females, we observe an initial sharp decrease in suicide ideation until age ~25 followed by a stabilization through middle adulthood. In males, the trajectory of suicide ideation followed a more stable decline across ages. We also found a significant sex effect with overall higher suicide ideation in females than in males.

Table 2.

Longitudinal Growth Model Fit in the Subset of Participants of European Ancestry (N=4905)

Models AIC BIC Log-likelihood Δ Chisq Δ df P
A. Growth trajectory of suicide ideation
Unconditional model 7852.3 7876.1 −3923.1 - - -
Linear growth model 7687.7 7719.6 −3839.9 - - -
Quadratic growth model 7657.3 7697.1 −3823.7 32.433 1 1.234e-08
Linear growth model with Sex*Age 7643.4 7691.2 −3815.7 15.872 1 6.779e-05
Quadratic growth model with Sex*Age 7599.6 7663.3 -3791.8 47.814 2 4.143e-11
PGS as a predictor of the growth trajectory
PGS as a fixed effect 7581.8 7653.4 −3781.9 19.809 1 8.556e-06
PGS*Sex 7581.9 7661.4 −3780.9 1.9286 1 0.1649
PGS*Age 7584.3 7671.8 −3781.1 0.0000 1 1.0000
PGS*Sex and PGS*Age 7584.4 7679.9 −3780.2 1.8714 1 0.1713
PGS*Sex*Age 7587.0 7698.4 −3779.5 1.3665 2 0.5050
B. Growth trajectory of suicide attempt
Unconditional model 5657.2 5677.0 −2825.6 - - -
Linear growth model 4003.2 4032.0 −1997.6 - - -
Quadratic growth model 3405.2 3441.3 −1697.6 599.95 1 <2e-16
Linear growth model with Sex*Age 3902.5 3945.8 −1945.3 0.00 1 1
Quadratic growth model with Sex*Age 3292.1 3349.7 −1638.0 614.48 2 <2e-16
PGS as a predictor of the growth trajectory
PGS as a fixed effect 3288.1 3352.9 −1635.0 5.9708 1 0.01454
PGS*Sex 3289.6 3361.6 −1634.8 0.4747 1 0.49083
PGS*Age 3291.4 3370.6 −1634.7 0.2057 1 0.65018
PGS*Sex and PGS*Age 3292.9 3379.4 −1634.5 0.4672 1 0.49429
PGS*Sex*Age 3293.0 3393.9 −1632.5 3.9154 2 0.14118
Open in a new tab

Notes: AIC = Akaike Information Criterion, BIC = Bayesian Information Criterion, Chisq = Chi-square test, DF = Degree of Freedom, P = P-values

Figure 1.

Figure 1.

Open in a new tab

The trajectories of Suicidal Thoughts and Behaviors (STB)

Figure 1 depicts the trajectories of (A) suicide ideation and (B) suicide attempt in participants of European Ancestry (EA) and (C) suicide ideation and (D) suicide attempt in participants of African Ancestry (AA). The Y-axis represents the mean level of STB and the X-axis represents age (from 12 to 40). The solid line denotes the slope and the shaded areas the 95% Confidence Intervals. Slopes for females are shown in blue and slopes for males are shown in red.

From this quadratic growth model, we tested additional models to explore the role of PGS for suicide ideation in change of ideation across ages. The best-fitting model included PGS as a fixed effect (Table 2.A.) and showed that higher PGS for suicide ideation was associated with a higher risk of suicide ideation at baseline but not with change over time (see Table S1 for full results).

AA group.

The ICC for suicide ideation was 0.162. Table 3.A. reports the fit statistics for the models evaluating the growth trajectories of suicide ideation. As for the EA group, the best-fitting model was a quadratic growth model including a sex-by-age interaction (AIC = 1284.0, BIC = 1338.5). As evidenced by this interaction term, we observed sex-specific trajectories for suicide ideation. In females, suicide ideation decreased sharply until age ~20, then stabilized between ages 20-30, and tended to increase after age 40. In males, we found a slight decrease in the first 5 years of adolescence (~ age 15) followed by an overall stabilization of suicide ideation from adolescence to middle adulthood (Table S2; Figure 1, part C).

Table 3.

Longitudinal Growth Model Fit in the Subset of Participants of African Ancestry (N=1654)

Models AIC BIC Log-likelihood Δ Chisq Δ df P
A. Growth trajectory of suicide ideation
Unconditional model 1332.2 1352.6 −663.1 - - -
Linear growth model 1325.5 1352.8 −658.77 - - -
Quadratic growth model 1308.6 1342.7 −649.32 18.9001 1 1.377e-05
Linear growth model with Sex*Age 1304.0 1345.0 −646.02 6.5871 1 0.01027
Quadratic growth model with Sex*Age 1284.0 1338.5 −634.00 24.0468 2 6.002e-06
PGS as a predictor of the growth trajectory
PGS as a fixed effect 1284.6 1345.9 −633.29 1.4206 1 0.233
PGS*Sex 1281.3 1349.5 −630.66 5.2555 1 0.02188
PGS*Age 1287.0 1362.0 −632.50 0.0000 1 1.0000
PGS*Sex and PGS*Age 1283.5 1365.3 −629.74 5.5205 1 0.01879
PGS*Sex*Age 1286.5 1381.9 −629.23 1.0178 2 0.60115
B. Growth trajectory of suicide attempt
Unconditional model 1846.1 1864.1 −920.1 - - -
Linear growth model 1570.2 1594.2 −781.09 - - -
Quadratic growth model 1347.0 1377.1 −668.52 225.14 1 <2e-16
Linear growth model with Sex*Age 1554.8 1590.8 −771.38 0.00 1 1
Quadratic growth model with Sex*Age 1341.7 1389.7 −662.84 217.08 2 <2e-16
PGS as a predictor of the growth trajectory
PGS as a fixed effect 1554.2 1596.2 −770.09 2.586 1 0.1078
PGS*Sex 1555.4 1603.5 −769.72 0.7498 1 0.3866
PGS*Age 1555.9 1603.9 −769.93 0.0000 0
PGS*Sex and PGS*Age 1557.1 1611.1 −769.56 0.7435 1 0.3885
PGS*Sex*Age 1559.1 1619.1 −769.53 0.0560 1 0.8130
Open in a new tab

Notes: AIC = Akaike Information Criterion, BIC = Bayesian Information Criterion, Chisq = Chi-square test, DF = Degree of Freedom, P = P-values. Best-fitting models are shown in bold.

From this model, we explored the role of PGS for suicide ideation. The model including a PGS-by-sex interaction fit the data best (Table 3.A.) and showed that higher PGS for suicide ideation was associated with a higher risk of suicide ideation in males only at baseline (Figure 2; Table S2). See Table S2 (supplement) for full results.

Figure 2.

Figure 2.

Open in a new tab

Aggregate genetic liability for suicide ideation in participants of African Ancestry

Figure 2 indicates the association between suicide ideation and aggregate genetic liability (polygenic score; PGS) for suicide ideation in participants of African Ancestry (AA) separately for females (blue) and males (red). Results show an effect of PGS in m ales only.

Analyses controlled for sampling weights are presented in Tables S1 (EA) and S2 (AA) and do not indicate substantial differences (Supplement).

Trajectories of suicide attempt

EA group.

From the unconditional model, we found an ICC of 0.092, suggesting some within-individual variance in suicide attempt over time. Table 2.B. shows that the quadratic growth model including a sex-by-age interaction fit the data best (AIC= 3292.1, BIC = 3349.7). Results from this model (Table S3; Figure 1, part B) indicated an overall decline in the rate of suicide attempt across adolescence into adulthood for both sexes. As evidenced by the significant age-by-sex interaction term, the shape of the decline varied by sex. Among females, the decline was more pronounced from adolescence until age ~25-30, and less steep from that point through middle adulthood. Among males, the decline was less pronounced across adulthood.

We tested additional models to explore the role of PGS. The best-fitting model included PGS as a fixed effect (Table 2.B.), showing overall that higher PGS was associated with a higher risk of suicide attempt at baseline, similar to the results for ideation in the EA sample (see Table S3 for full results).

AA group.

In participants of AA, we found an ICC of 0.077 for suicide attempt. A quadratic growth model including a sex-by-age interaction suited the data best (AIC= 1341.7, BIC = 1389.7, see Table 3.B.). Suicide attempt was characterized by a steep decrease until age ~25, a stabilization, and a tendency to increase after age ~35 in both sexes, with a sharper initial decrease in females. However, after age ~30, the slopes of females and males crossed and 95% CIs overlapped (Table S4; Figure 1, part D).

We tested additional models to explore the role of PGS, but the results suggested that adding PGS as a predictor did not improve the model fit (Table 3.B., Table S4).

Analyses controlled for sampling weights are presented in Tables S3 (EA) and S4 (AA); results did not differ substantively (Supplement).

Discussion

This study evaluates the changes in STB over time in a population-based cohort from the US. Our results show an overall decrease of STB from adolescence to the beginning of adulthood with important sex differences: In females, we observed higher initial levels of STB and sharper decreases across ages. Adulthood was then characterized by low STB, with a possible increase after age 35. Aggregate genetic liability was related to the baseline levels of STB but not its change over time. These findings provide insight into the time periods related to high and low STB risks across sexes in participants of EA and AA.

Findings are consistent with prior studies (Curtin, 2020; Hawton et al., 2012) indicating that adolescence is a high-risk period for both suicide ideation and suicide attempt. Considering the trajectories of STB, we further highlight an important decrease in STB from adolescence to the beginning of adulthood (age 20). This period of adolescent development is related to brain maturation, in which heterochronicity between limbic and prefrontal systems may be associated with risky behaviors and depression (Lannoy & Sullivan, 2021; Steinberg, 2008). The limbic brain system matures early in adolescence, which leads to greater emotionality and a higher sensitivity of the serotoninergic system (Morgane et al., 2005). In contrast, the prefrontal system matures gradually until the beginning of adulthood (Giedd, 2008), leading progressively to lower impulsivity, emotionality, and sensation seeking (Steinberg, 2008). Ongoing brain maturation and hormonal changes could contribute to the high but declining risk of STB during adolescence (Manceaux et al., 2015; Schwartz et al., 2019), though additional explorations (which are outside the scope of the current research questions) are necessary to precisely determine how maturation is related to STB.

The ancestry-specific analyses highlight distinct patterns of change: The rate of STB was higher during adolescence than adulthood in both EA and AA participants, but the risk appeared to stabilize in EA while it tended to increase in adulthood in AA. Though the prevalence was still lower than in adolescence, our results suggest that another risk period may occur after age 35 in participants of AA. This potential increase was also observed in participants of EA, but to a lesser extent. Additional follow-ups are important to characterize this increased risk in both groups. Our findings also support the importance of taking into account sex when describing STB trajectories, as all best-fitting models included a sex-by-age interaction. In line with previous studies (Miranda-Mendizabal et al., 2019), in adolescence, females presented higher STB than males. However, the present study indicated that at the beginning of adulthood, the differences between males and females attenuated. To our knowledge, there is no prior longitudinal study documenting sex differences across developmental periods. This observation reinforces the importance of examining the trajectories of STB over time to complement other statistical methods investigating the heterogeneity of STB. As maturation might explain the decline in STB risk across adolescence, sex differences in STB might also be related to distinct maturation in girls and boys (Chaku & Hoyt, 2019; Lannoy, Pfefferbaum, et al., 2022).

Finally, these findings inform our understanding of the role of aggregate genetic liability in the trajectories of STB. In line with previous studies using PGS (Barzilay et al., 2022; A. R. Docherty et al., 2023; Lannoy, Mars, et al., 2022; Lee et al., 2022), the overall effect size of PGS is small. However, we found substantial improvements in model fit when adding PGS as a predictor of the growth trajectories (except for suicide attempt in the AA group), providing further support for genetic liability as an important indicator of STB risk. Interestingly, findings suggest that one’s genetic liability was related to early risk of STB (baseline level) but was less informative of later changes in risk. This is somewhat consistent with results from twin studies (Edwards et al., 2021), indicating a more prominent role of genetic factors in young people. These results could also be understood through the lens of ideation-to-action theories (Klonsky & May, 2015; O'Connor & Kirtley, 2018), which highlight genetic liability as a pre-dispositional factor and suggest that other risk factors and personal traits (e.g., suicide capability) contribute to the change of risk over time. However, how STB trajectories map onto existing theoretical models warrants further investigation. In the AA sample, we found that the effect of PGS varied by sex, with a role of PGS for suicide ideation in males only. However, this observation warrants replication using GWAS summary statistics from a large sample of comparable ancestry, as those used to compute PGS in this study were derived from a veteran sample that was largely males of EA (Ashley-Koch et al., 2023), and the performance of PGS across ancestry groups is diminished relative to within an ancestry group (Kachuri et al., 2024).

Utilizing growth modeling, the strengths of this study are to (i) show population-average changes of STB across ages, (ii) demonstrate the specific growth pattern in males and females in different ancestry groups (participants of EA and AA), and (iii) document the role of genetic liability in predicting baseline levels of STB but not their changes over time. These findings also need to be considered in the context of limitations. First, the evaluation of suicide attempt in Add Health was conditioned on suicide ideation in Waves I-III but not in Waves IV and V. In line with recent results (Bryan et al., 2020; Padron et al., 2025), assessments in the last two waves of data collection allowed us to capture suicide attempts that were not preceded by ideation. We believe this is an important improvement and wanted to consider it in our analyses, though this might influence the stability of our results at later ages and our modeling approach. Indeed, the current results presented growth models with random intercept (i.e., allowing participants to vary in their baseline level of STB). Models including both random intercept and slopes (i.e., allowing participants to vary over time) could not be fitted for suicide attempt due to the data structure. Sensitivity analyses were conducted for suicide ideation and indicated highly similar fixed-effect estimates, suggesting that conclusions were robust to the random-effects specification. However, these results require replication. Second, though our results were supported in several sensitivity analyses, our statistical approach did not fully handle missing data. Third, while it is important to pursue ancestry-specific analyses, suicide attempt is relatively uncommon, and we have reduced statistical power to evaluate its change over time in the AA group, particularly at later ages. In addition, current GWAS summary statistics are derived primarily from samples of EA. Though we used PRS-CSx and ancestry-specific sumstats, the smaller AA discovery sample sizes may result in reduced statistical power. It is also worth noting that, although the inclusion of PGS substantially improved the model fit, the current predictive utility of PGS is low (A. Docherty et al., 2021). Fourth, this work provides important information on STB changes over time across sex and ancestry groups, but the results need to be considered in the context of previously documented risk and protective factors. For example, our findings indicate a lower risk of STB among participants of African ancestry, whereas prior studies using self-reported race have suggested that the lower prevalence of STB among self-identified Black participants may be largely attributable to protective factors such as higher levels of religious and familial involvement (Davidson & Wingate, 2011). Finally, data collection for the Add Health sample started in 1994, and cohort differences in the rates of STB across ancestry and sex could reduce the generalizability of our findings.

To conclude, the current study evaluates the trajectory of STB from adolescence to adulthood in two ancestry groups by considering the roles of sex and aggregate genetic liability. Findings show adolescence as the highest risk period for both suicide ideation and attempt. This reinforces the need to implement prevention starting early in adolescence, with specific attention to females and those with high genetic liability (i.e., family history of suicidal behavior could be used as a proxy to estimate genetic risk). It also calls for a better understanding of the mechanisms that would explain this high risk during adolescence. Promising avenues are related to the association between brain maturation, hormonal changes, and STB across ages (Ho et al., 2022; Manceaux et al., 2015). Finally, we observed a potential increase in STB risk after age 35, particularly in the AA group. Longer follow-ups would be important to characterize this as a second period of risk.

Supplementary Material

Supplement

Highlights.

  • We evaluate the trajectories of suicide ideation/attempt from 12-40 across ancestry

  • The best-fitting models included sex and showed the role of genetic liability

  • Ideation/attempt decrease from adolescence to adulthood and increase after age 35

Acknowledgments

This research uses data from Add Health, funded by grant P01 HD31921 (Harris) from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD), with cooperative funding from 23 other federal agencies and foundations. Add Health is currently directed by Robert A. Hummer and funded by the National Institute on Aging cooperative agreements U01 AG071448 (Hummer) and U01AG071450 (Aiello and Hummer) at the University of North Carolina at Chapel Hill. Add Health was designed by J. Richard Udry, Peter S. Bearman, and Kathleen Mullan Harris at the University of North Carolina at Chapel Hill.

This work was also supported by the National Institutes of Health: NIMH to AE (R01MH129356) and NIAAA to SL (K99AA030611). The content is solely the responsibility of the authors and does not necessarily represent the official views of the funders.

Footnotes

Declaration of interest statement: None.

References

  1. Ashley-Koch AE, Kimbrel NA, Qin XJ, Lindquist JH, Garrett ME, Dennis MF, Hair LP, Huffman JE, Jacobson DA, Madduri RK, Coon H, Docherty AR, Kang J, Mullins N, Ruderfer DM, Program VAMV, Workgroup MVPSE, International Suicide Genetics C, Harvey PD, McMahon BH, Oslin DW, Hauser ER, Hauser MA, & Beckham JC (2023). Genome-wide association study identifies four pan-ancestry loci for suicidal ideation in the Million Veteran Program. PLoS Genet, 19(3), e1010623. 10.1371/journal.pgen.1010623 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Barr PB, Mallard TT, Sanchez-Roige S, Poore HE, Linner RK, Collaborators C, Waldman ID, Palmer AA, Harden KP, & Dick DM (2022). Parsing genetically influenced risk pathways: genetic loci impact problematic alcohol use via externalizing and specific risk. Transl Psychiatry, 12(1), 420. 10.1038/s41398-022-02171-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Barzilay R, Visoki E, Schultz LM, Warrier V, Daskalakis NP, & Almasy L (2022). Genetic risk, parental history, and suicide attempts in a diverse sample of US adolescents. Front Psychiatry, 13, 941772. 10.3389/fpsyt.2022.941772 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bornheimer LA, Wang K, Zhang A, Li J, Trim EE, Ilgen M, & King CA (2022). National trends in non-fatal suicidal behaviors among adults in the USA from 2009 to 2017. Psychol Med, 52(6), 1031–1039. 10.1017/S0033291720002755 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Braudt DB, & Harris MK (2018). Polygenic Scores (PGSs) in the National Longitudinal Study of Adolescent to Adult Health (Add Health) – Release 1. 2018. 10.17615/9g92-vc17 [DOI] [Google Scholar]
  6. Bryan CJ, Butner JE, May AM, Rugo KF, Harris J, Oakey DN, Rozek DC, & Bryan AO (2020). Nonlinear change processes and the emergence of suicidal behavior: a conceptual model based on the fluid vulnerability theory of suicide. New Ideas Psychol, 57. 10.1016/j.newideapsych.2019.100758 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Carter SP, Campbell SB, Wee JY, Law KC, Lehavot K, Simpson T, & Reger MA (2022). Suicide Attempts Among Racial and Ethnic Groups in a Nationally Representative Sample. J Racial Ethn Health Disparities, 9(5), 1783–1793. 10.1007/s40615-021-01115-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Casey BJ, Jones RM, & Hare TA (2008). The Adolescent Brain. Ann N Y Acad Sci, 1124, 111–126. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Centers for Disease Control and Prevention. (2025). Suicide data and statistics https://www.cdc.gov/suicide/facts/data.html
  10. Chaku N, & Hoyt LT (2019). Developmental Trajectories of Executive Functioning and Puberty in Boys and Girls. J Youth Adolesc, 48(7), 1365–1378. 10.1007/s10964-019-01021-2 [DOI] [PubMed] [Google Scholar]
  11. Chang CC, Chow CC, Tellier LC, Vattikuti S, Purcell SM, & Lee JJ (2015). Second-generation PLINK: rising to the challenge of larger and richer datasets. Gigascience, 4, 7. 10.1186/s13742-015-0047-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Committee on the Use of Race, E., and Ancestry as Population Descriptors in Genomics Research. (2023). Using Population Descriptors in Genetics and Genomics Research: A New Framework for an Evolving Field. National Academies Sciences Engineering Medicine. [PubMed] [Google Scholar]
  13. Curtin SC (2020). State Suicide Rates Among Adolescents and Young Adults Aged 10–24: United States, 2000–2018 ( [PubMed] [Google Scholar]
  14. Czyz EK, & King CA (2015). Longitudinal trajectories of suicidal ideation and subsequent suicide attempts among adolescent inpatients. J Clin Child Adolesc Psychol, 44(1), 181–193. 10.1080/15374416.2013.836454 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Davidson CL, & Wingate LR (2011). Racial Disparities in Risk and Protective Factors for Suicide. Journal of Black Psychology, 37(4), 499–516. 10.1177/0095798410397543 [DOI] [Google Scholar]
  16. Docherty A, Kious B, Brown T, Francis L, Stark L, Keeshin B, Botkin J, DiBlasi E, Gray D, & Coon H (2021). Ethical concerns relating to genetic risk scores for suicide. Am J Med Genet B Neuropsychiatr Genet, 186(8), 433–444. 10.1002/ajmg.b.32871 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Docherty AR, Mullins N, Ashley-Koch AE, Qin X, Coleman JRI, Shabalin A, Kang J, Murnyak B, Wendt F, Adams M, Campos AI, DiBlasi E, Fullerton JM, Kranzler HR, Bakian AV, Monson ET, Renteria ME, Walss-Bass C, Andreassen OA, Behera C, Bulik CM, Edenberg HJ, Kessler RC, Mann JJ, Nurnberger JI Jr., Pistis G, Streit F, Ursano RJ, Polimanti R, Dennis M, Garrett M, Hair L, Harvey P, Hauser ER, Hauser MA, Huffman J, Jacobson D, Madduri R, McMahon B, Oslin DW, Trafton J, Awasthi S, Berrettini WH, Bohus M, Chang X, Chen HC, Chen WJ, Christensen ED, Crow S, Duriez P, Edwards AC, Fernandez-Aranda F, Galfalvy H, Gandal M, Gorwood P, Guo Y, Hafferty JD, Hakonarson H, Halmi KA, Hishimoto A, Jain S, Jamain S, Jimenez-Murcia S, Johnson C, Kaplan AS, Kaye WH, Keel PK, Kennedy JL, Kim M, Klump KL, Levey DF, Li D, Liao SC, Lieb K, Lilenfeld L, Marshall CR, Mitchell JE, Okazaki S, Otsuka I, Pinto D, Powers A, Ramoz N, Ripke S, Roepke S, Rozanov V, Scherer SW, Schmahl C, Sokolowski M, Starnawska A, Strober M, Su MH, Thornton LM, Treasure J, Ware EB, Watson HJ, Witt SH, Woodside DB, Yilmaz Z, Zillich L, Adolfsson R, Agartz I, Alda M, Alfredsson L, Appadurai V, Artigas MS, Van der Auwera S, Azevedo MH, Bass N, Bau CHD, Baune BT, Bellivier F, Berger K, Biernacka JM, Bigdeli TB, Binder EB, Boehnke M, Boks MP, Braff DL, Bryant R, Budde M, Byrne EM, Cahn W, Castelao E, Cervilla JA, Chaumette B, Corvin A, Craddock N, Djurovic S, Foo JC, Forstner AJ, Frye M, Gatt JM, Giegling I, Grabe HJ, Green MJ, Grevet EH, Grigoroiu-Serbanescu M, Gutierrez B, Guzman-Parra J, Hamshere ML, Hartmann AM, Hauser J, Heilmann-Heimbach S, Hoffmann P, Ising M, Jones I, Jones LA, Jonsson L, Kahn RS, Kelsoe JR, Kendler KS, Kloiber S, Koenen KC, Kogevinas M, Krebs MO, Landen M, Leboyer M, Lee PH, Levinson DF, Liao C, Lissowska J, Mayoral F, McElroy SL, McGrath P, McGuffin P, McQuillin A, Mehta D, Melle I, Mitchell PB, Molina E, Morken G, Nievergelt C, Nothen MM, O'Donovan MC, Ophoff RA, Owen MJ, Pato C, Pato MT, Penninx B, Potash JB, Power RA, Preisig M, Quested D, Ramos-Quiroga JA, Reif A, Ribases M, Richarte V, Rietschel M, Rivera M, Roberts A, Roberts G, Rouleau GA, Rovaris DL, Sanders AR, Schofield PR, Schulze TG, Scott LJ, Serretti A, Shi J, Sirignano L, Sklar P, Smeland OB, Smoller JW, Sonuga-Barke EJS, Trzaskowski M, Tsuang MT, Turecki G, Vilar-Ribo L, Vincent JB, Volzke H, Walters JTR, Weickert CS, Weickert TW, Weissman MM, Williams LM, Wray NR, Zai CC, Agerbo E, Borglum AD, Breen G, Demontis D, Erlangsen A, Gelernter J, Glatt SJ, Hougaard DM, Hwu HG, Kuo PH, Lewis CM, Li QS, Liu CM, Martin NG, McIntosh AM, Medland SE, Mors O, Nordentoft M, Olsen CM, Porteous D, Smith DJ, Stahl EA, Stein MB, Wasserman D, Werge T, Whiteman DC, Willour V, Program, V. A. M. V., Workgroup, M. V. P. S. E., Suicide Working Group of the Psychiatric Genomics, C., Major Depressive Disorder Working Group of the Psychiatric Genomics, C., Bipolar Disorder Working Group of the Psychiatric Genomics, C., Schizophrenia Working Group of the Psychiatric Genomics, C., Eating Disorder Working Group of the Psychiatric Genomics, C., German Borderline Genomics, C., Coon H, Beckham JC, Kimbrel NA, & Ruderfer DM (2023). GWAS Meta-Analysis of Suicide Attempt: Identification of 12 Genome-Wide Significant Loci and Implication of Genetic Risks for Specific Health Factors. American Journal of Psychiatry, 180(10), 723–738. 10.1176/appi.ajp.21121266 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Edwards AC, Heron J, Vladimirov V, Wolen AR, Adkins DE, Aliev F, Hickman M, & Kendler KS (2017). The Rate of Change in Alcohol Misuse Across Adolescence is Heritable. Alcohol Clin Exp Res, 41(1), 57–64. 10.1111/acer.13262 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Edwards AC, Ohlsson H, Moscicki E, Crump C, Sundquist J, Lichtenstein P, Kendler KS, & Sundquist K (2021). On the Genetic and Environmental Relationship Between Suicide Attempt and Death by Suicide. American Journal of Psychiatry, 178(11), 1060–1069. 10.1176/appi.ajp.2020.20121705 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Erausquin JT, McCoy TP, Bartlett R, & Park E (2019). Trajectories of Suicide Ideation and Attempts from Early Adolescence to Mid-Adulthood: Associations with Race/Ethnicity. J Youth Adolesc, 48(9), 1796–1805. 10.1007/s10964-019-01074-3 [DOI] [PubMed] [Google Scholar]
  21. Ge T, Irvin MR, Patki A, Srinivasasainagendra V, Lin YF, Tiwari HK, Armstrong ND, Benoit B, Chen CY, Choi KW, Cimino JJ, Davis BH, Dikilitas O, Etheridge B, Feng YA, Gainer V, Huang H, Jarvik GP, Kachulis C, Kenny EE, Khan A, Kiryluk K, Kottyan L, Kullo IJ, Lange C, Lennon N, Leong A, Malolepsza E, Miles AD, Murphy S, Namjou B, Narayan R, O'Connor MJ, Pacheco JA, Perez E, Rasmussen-Torvik LJ, Rosenthal EA, Schaid D, Stamou M, Udler MS, Wei WQ, Weiss ST, Ng MCY, Smoller JW, Lebo MS, Meigs JB, Limdi NA, & Karlson EW (2022). Development and validation of a trans-ancestry polygenic risk score for type 2 diabetes in diverse populations. Genome Med, 14(1), 70. 10.1186/s13073-022-01074-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Genomes Project C, Auton A, Brooks LD, Durbin RM, Garrison EP, Kang HM, Korbel JO, Marchini JL, McCarthy S, McVean GA, & Abecasis GR (2015). A global reference for human genetic variation. Nature, 526(7571), 68–74. 10.1038/nature15393 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Geoffroy MC, Bouchard S, Per M, Khoury B, Chartrand E, Renaud J, Turecki G, Colman I, & Orri M (2022). Prevalence of suicidal ideation and self-harm behaviours in children aged 12 years and younger: a systematic review and meta-analysis. Lancet Psychiatry, 9(9), 703–714. 10.1016/S2215-0366(22)00193-6 [DOI] [PubMed] [Google Scholar]
  24. Geoffroy MC, Orri M, Girard A, Perret LC, & Turecki G (2021). Trajectories of suicide attempts from early adolescence to emerging adulthood: prospective 11-year follow-up of a Canadian cohort. Psychol Med, 51(11), 1933–1943. 10.1017/S0033291720000732 [DOI] [PubMed] [Google Scholar]
  25. Giedd JN (2008). The teen brain: insights from neuroimaging. J Adolesc Health, 42(4), 335–343. 10.1016/j.jadohealth.2008.01.007 [DOI] [PubMed] [Google Scholar]
  26. Harris K. (2013). The Add Health Study: Design and Accomplishments. University of North Carolina at Chapel Hill; https://addhealth.cpc.unc.edu/wp-content/uploads/docs/user_guides/DesignPaperWave_I-IV.pdf [Google Scholar]
  27. Harris KM, Halpern CT, Whitsel EA, Hussey JM, Tabor J, Entzel PP, & Udry JR (2009). The National Longitudinal Study of Adolescent Health: Research Design. Carolina Population Center, University of North Carolina at Chapel Hill. https://addhealth.cpc.unc.edu/documentation/study-design/ [Google Scholar]
  28. Harris KM, & Udry JR (2022). National Longitudinal Study of Adolescent to Adult Health (Add Health), 1994-2018 [Public Use].Carolina Population Center, University of North Carolina-Chapel Hill [distributor], Inter-university Consortium for Political and Social Research [distributor]. 10.3886/ICPSR21600.v25 [DOI] [Google Scholar]
  29. Hawton K, Saunders KE, & O'Connor RC (2012). Self-harm and suicide in adolescents. Lancet, 379(9834), 2373–2382. 10.1016/S0140-6736(12)60322-5 [DOI] [PubMed] [Google Scholar]
  30. Highland HM, Avery CL, Duan Q, Li Y, & Harris KM (2018). Quality control analysis of Add Health GWAS data https://addhealth.cpc.unc.edu/wp-content/uploads/docs/user_guides/AH_GWAS_QC.pdf [Google Scholar]
  31. Ho TC, Gifuni AJ, & Gotlib IH (2022). Psychobiological risk factors for suicidal thoughts and behaviors in adolescence: a consideration of the role of puberty. Mol Psychiatry, 27(1), 606–623. 10.1038/s41380-021-01171-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Ivey-Stephenson AZ, Crosby AE, Hoenig JM, Gyawali S, Park-Lee E, & Hedden SL (2022). Suicidal Thoughts and Behaviors Among Adults Aged ≥18 Years —United States, 2015–2019. M. M. M. W. Rep [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Ivey-Stephenson AZ, Demissie Z, Crosby AE, Stone DM, Gaylor E, Wilkins NJ, Lowry R, & Brown M (2020). Suicidal Ideation and Behaviors Among High School Students — Youth Risk Behavior Survey, United States, 2019. MMWR [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Kachuri L, Chatterjee N, Hirbo J, Schaid DJ, Martin I, Kullo IJ, Kenny EE, Pasaniuc B, Polygenic Risk Methods in Diverse Populations Consortium Methods Working, G., Witte JS, & Ge T (2024). Principles and methods for transferring polygenic risk scores across global populations. Nat Rev Genet, 25(1), 8–25. 10.1038/s41576-023-00637-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Kimbrel NA, Ashley-Koch AE, Qin XJ, Lindquist JH, Garrett ME, Dennis MF, Hair LP, Huffman JE, Jacobson DA, Madduri RK, Trafton JA, Coon H, Docherty AR, Kang J, Mullins N, Ruderfer DM, Program VAMV, Workgroup MVPSE, International Suicide Genetics C, Harvey PD, McMahon BH, Oslin DW, Hauser ER, Hauser MA, & Beckham JC (2022). A genome-wide association study of suicide attempts in the million veterans program identifies evidence of pan-ancestry and ancestry-specific risk loci. Mol Psychiatry, 27(4), 2264–2272. 10.1038/s41380-022-01472-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Klonsky ED, & May AM (2015). The Three-Step Theory (3ST): A New Theory of Suicide Rooted in the “Ideation-to-Action” Framework. International Journal of Cognitive Therapy, 8(2), 114–129. [Google Scholar]
  37. Lannoy S, Mars B, Heron J, & Edwards AC (2022). Suicidal ideation during adolescence: The roles of aggregate genetic liability for suicide attempts and negative life events in the past year. J Child Psychol Psychiatry, 63(10), 1164–1173. 10.1111/jcpp.13653 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Lannoy S, Pfefferbaum A, Le Berre AP, Thompson WK, Brumback T, Schulte T, Pohl KM, De Bellis MD, Nooner KB, Baker FC, Prouty D, Colrain IM, Nagel BJ, Brown SA, Clark DB, Tapert SF, Sullivan EV, & Muller-Oehring EM (2022). Growth trajectories of cognitive and motor control in adolescence: How much is development and how much is practice? Neuropsychology, 36(1), 44–54. 10.1037/neu0000771 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Lannoy S, & Sullivan EV (2021). Trajectories of brain development reveal times of risk and factors promoting resilience to alcohol use during adolescence. Int Rev Neurobiol, 160, 85–116. 10.1016/bs.irn.2021.08.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Lee PH, Doyle AE, Silberstein M, Jung JY, Liu RT, Perlis RH, Roffman J, Smoller JW, Fava M, & Kessler RC (2022). Associations Between Genetic Risk for Adult Suicide Attempt and Suicidal Behaviors in Young Children in the US. JAMA Psychiatry, 79(10), 971–980. 10.1001/jamapsychiatry.2022.2379 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Liu RT, Walsh RFL, Sheehan AE, Cheek SM, & Sanzari CM (2022). Prevalence and Correlates of Suicide and Nonsuicidal Self-injury in Children: A Systematic Review and Meta-analysis. JAMA Psychiatry, 79(7), 718–726. 10.1001/jamapsychiatry.2022.1256 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Lopez R, Harris KM, Seibel L, Thomas SC, Leichtweis RN, & Esposito-Smythers C (2023). Trajectories of adolescent suicidal ideation and depressive symptoms during partial hospitalization: Clinical and demographic characteristics as predictors of change. Psychol Serv. 10.1037/ser0000796 [DOI] [PubMed] [Google Scholar]
  43. Madhavan S, Olino TM, Klein DN, & Seeley JR (2021). Longitudinal predictors of suicidal ideation: Emerging to early adulthood. J Psychiatr Res, 142, 210–217. 10.1016/j.jpsychires.2021.08.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Manceaux P, Jacques D, & Zdanowicz N (2015). Hormonal and developmental influences on adolescent suicide: A systematic review. Psychiatria Danubima, 27, 300–304. [PubMed] [Google Scholar]
  45. Miranda-Mendizabal A, Castellvi P, Pares-Badell O, Alayo I, Almenara J, Alonso I, Blasco MJ, Cebria A, Gabilondo A, Gili M, Lagares C, Piqueras JA, Rodriguez-Jimenez T, Rodriguez-Marin J, Roca M, Soto-Sanz V, Vilagut G, & Alonso J (2019). Gender differences in suicidal behavior in adolescents and young adults: systematic review and meta-analysis of longitudinal studies. International Journal of Public Health, 64(2), 265–283. 10.1007/s00038-018-1196-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Morgane PJ, Galler JR, & Mokler DJ (2005). A review of systems and networks of the limbic forebrain/limbic midbrain. Prog Neurobiol, 75(2), 143–160. 10.1016/j.pneurobio.2005.01.001 [DOI] [PubMed] [Google Scholar]
  47. Mullins N, Kang J, Campos AI, Coleman JRI, Edwards AC, Galfalvy H, Levey DF, Lori A, Shabalin A, Starnawska A, Su M-H, Watson HJ, Adams M, Awasthi S, Gandal M, Hafferty JD, Hishimoto A, Kim M, Okazaki S, Otsuka I, Ripke S, Ware EB, Bergen AW, Berrettini WH, Bohus M, Brandt H, Chang X, Chen WJ, Chen H-C, Crawford S, Crow S, DiBlasi E, Duriez P, Fernández-Aranda F, Fichter MM, Gallinger S, Glatt SJ, Gorwood P, Guo Y, Hakonarson H, Halmi KA, Hwu H-G, Jain S, Jamain S, Jiménez-Murcia S, Johnson C, Kaplan AS, Kaye WH, Keel PK, Kennedy JL, Klump KL, Levitan RD, Li D, Liao S-C, Lieb K, Lilenfeld L, Liu C-M, Magistretti PJ, Marshall CR, Mitchell JE, Monson ET, Myers RM, Pinto D, Powers A, Ramoz N, Roepke S, Rotondo A, Rozanov V, Scherer SW, Schmahl C, Sokolowski M, Strober M, Thornton LM, Treasure J, Tsuang MT, Via MCL, Witt SH, Woodside DB, Yilmaz Z, Zillich L, Adolfsson R, Agartz I, Air TM, Alda M, Alfredsson L, Andreassen OA, Anjorin A, Appadurai V, Artigas MS, Van der Auwera S, Azevedo MH, Bass N, Bau CHD, Baune BT, Bellivier F, Berger K, Biernacka JM, Bigdeli TB, Binder EB, Boehnke M, Boks M, Bosch R, Braff DL, Bryant R, Budde M, Byrne EM, Cahn W, Casas M, Castelao E, Cervilla JA, Chaumette B, Cichon S, Corvin A, Craddock N, Craig D, Degenhardt F, Djurovic S, Edenberg HJ, Fanous AH, Foo JC, Forstner AJ, Frye M, Fullerton JM, Gatt JM, Gejman PV, Giegling I, Grabe HJ, Green MJ, Grevet EH, Grigoroiu-Serbanescu M, Gutierrez B, Guzman-Parra J, Hamilton SP, Hamshere ML, Hartmann A, Hauser J, Heilmann-Heimbach S, Hoffmann P, Ising M, Jones I, Jones LA, Jonsson L, Kahn RS, Kelsoe JR, Kendler KS, Kloiber S, Koenen KC, Kogevinas M, Konte B, Krebs M-O, Landén M, Lawrence J, Leboyer M, Lee PH, Levinson DF, Liao C, Lissowska J, Lucae S, Mayoral F, McElroy SL, McGrath P, McGuffin P, McQuillin A, Medland S, Mehta D, Melle I, Milaneschi Y, Mitchell PB, Molina E, Morken G, Mortensen PB, Müller-Myhsok B, Nievergelt C, Nimgaonkar V, Nöthen MM, O’Donovan MC, Ophoff RA, Owen MJ, Pato C, Pato MT, Penninx BWJH, Pimm J, Pistis G, Potash JB, Power RA, Preisig M, Quested D, Ramos-Quiroga JA, Reif A, Ribasés M, Richarte V, Rietschel M, Rivera M, Roberts A, Roberts G, Rouleau GA, Rovaris DL, Rujescu D, Sánchez-Mora C, Sanders AR, Schofield PR, Schulze TG, Scott LJ, Serretti A, Shi J, Shyn SI, Sirignano L, Sklar P, Smeland OB, Smoller JW, Sonuga-Barke EJS, Spalletta G, Strauss JS, Świątkowska B, Trzaskowski M, Turecki G, Vilar-Ribó L, Vincent JB, Völzke H, Walters JTR, Weickert CS, Weickert TW, Weissman MM, Williams LM, Wray NR, Zai C, Agerbo E, Børglum AD, Breen G, Erlangsen A, Esko T, Gelernter J, Hougaard DM, Kessler RC, Kranzler HR, Li QS, Martin NG, McIntosh AM, Medland SE, Mors O, Nordentoft M, Olsen CM, Porteous D, Ursano RJ, Wasserman D, Werge T, Whiteman DC, Bulik CM, Coon H, Demontis D, Docherty AR, Kuo P-H, Lewis CM, Mann JJ, Rentería ME, Smith DJ, Stahl EA, Stein MB, Streit F, Willour V, & Ruderfer DM (2022). Dissecting the shared genetic architecture of suicide attempt, psychiatric disorders and known risk factors. Biological Psychiatry, 91(3), 313–327. 10.1101/2020.12.01.20241281 [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Nakar O, Brunner R, Schilling O, Chanen A, Fischer G, Parzer P, Carli V, Wasserman D, Sarchiapone M, Wasserman C, Hoven CW, Resch F, & Kaess M (2016). Developmental trajectories of self-injurious behavior, suicidal behavior and substance misuse and their association with adolescent borderline personality pathology. J Affect Disord, 197, 231–238. 10.1016/j.jad.2016.03.029 [DOI] [PubMed] [Google Scholar]
  49. Nkansah-Amankra S. (2013). Adolescent suicidal trajectories through young adulthood: prospective assessment of religiosity and psychosocial factors among a population-based sample in the United States. Suicide Life Threat Behav, 43(4), 439–459. 10.1111/sltb.12029 [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. O'Connor RC, & Kirtley OJ (2018). The integrated motivational-volitional model of suicidal behaviour. Philosophical Transactions of the Royal Society B: Biological Sciences 373(1754). 10.1098/rstb.2017.0268 [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Padron M, Liu L, & Pettit JW (2025). Suicide attempt in the absence of suicide ideation: Prevalence and correlates among youth detained in the juvenile legal system. J Affect Disord, 391, 119927. 10.1016/j.jad.2025.119927 [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Pereira JL, de Souza CA, Neyra JEM, Leite J, Cerqueira A, Mingroni-Netto RC, Soler JMP, Rogero MM, Sarti FM, & Fisberg RM (2024). Genetic Ancestry and Self-Reported "Skin Color/Race" in the Urban Admixed Population of Sao Paulo City, Brazil. Genes (Basel), 15(7). 10.3390/genes15070917 [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Peterson RE, Kuchenbaecker K, Walters RK, Chen CY, Popejoy AB, Periyasamy S, Lam M, Iyegbe C, Strawbridge RJ, Brick L, Carey CE, Martin AR, Meyers JL, Su J, Chen J, Edwards AC, Kalungi A, Koen N, Majara L, Schwarz E, Smoller JW, Stahl EA, Sullivan PF, Vassos E, Mowry B, Prieto ML, Cuellar-Barboza A, Bigdeli TB, Edenberg HJ, Huang H, & Duncan LE (2019). Genome-wide Association Studies in Ancestrally Diverse Populations: Opportunities, Methods, Pitfalls, and Recommendations. Cell, 179(3), 589–603. 10.1016/j.cell.2019.08.051 [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Prabhakar A, & Rice SE (2023). White House Report on Mental Health Research Priorities. The White House [Google Scholar]
  55. Prinstein MJ, Nock MK, Simon V, Aikins JW, Cheah CS, & Spirito A (2008). Longitudinal trajectories and predictors of adolescent suicidal ideation and attempts following inpatient hospitalization. J Consult Clin Psychol, 76(1), 92–103. 10.1037/0022-006X.76.1.92 [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. R Core Development Team. (2021). A language and environment for statistical computing. R Foundation for Statistical Computing. [Google Scholar]
  57. Ram N, & Grimm KJ (2009). Growth Mixture Modeling: A Method for Identifying Differences in Longitudinal Change Among Unobserved Groups. Int J Behav Dev, 33(6), 565–576. 10.1177/0165025409343765 [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Ruan Y, Lin YF, Feng YA, Chen CY, Lam M, Guo Z, Stanley Global Asia I, He L, Sawa A, Martin AR, Qin S, Huang H, & Ge T (2022). Improving polygenic prediction in ancestrally diverse populations. Nat Genet, 54(5), 573–580. 10.1038/s41588-022-01054-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Schwartz J, Ordaz SJ, Ho TC, & Gotlib IH (2019). Longitudinal decreases in suicidal ideation are associated with increases in salience network coherence in depressed adolescents. J Affect Disord, 245, 545–552. 10.1016/j.jad.2018.11.009 [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Shi X, Jiang L, Chen X, & Zhu Y (2022). Distinct trajectories of suicidal behaviors throughout the university stage and associated risk and protective factors: A large-scale prospective study. J Affect Disord, 319, 407–415. 10.1016/j.jad.2022.09.107 [DOI] [PubMed] [Google Scholar]
  61. Singer JD, & Willett JB (2003). Applied longitudinal Data Analysis: Modeling Change and Event Occurrence. Oxford University Press. [Google Scholar]
  62. Steinberg L (2008). A Social Neuroscience Perspective on Adolescent Risk-Taking. Dev Rev, 28(1), 78–106. 10.1016/j.dr.2007.08.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  63. Steinberg L, Albert D, Cauffman E, Banich M, Graham S, & Woolard J (2008). Age differences in sensation seeking and impulsivity as indexed by behavior and self-report: evidence for a dual systems model. Dev Psychol, 44(6), 1764–1778. 10.1037/a0012955 [DOI] [PubMed] [Google Scholar]
  64. Substance Abuse and Mental Health Services Administration. (2025). Key substance use and mental health indicators in the United States: Results from the 2024 National Survey on Drug Use and Health (HHS Publication No. PEP25-07-007, NSDUH Series H-60). Center for Behavioral Health Statistics and Quality, Substance Abuse and Mental Health Services Administration; https://www.samhsa.gov/data/report/2024-nsduh-annual-national-report [Google Scholar]
  65. Tang H, Quertermous T, Rodriguez S, Kardia SLR, Zhu X, Brown A, Pankow JS, Province MA, Hunt SC, Boerwinkle E, Schork NJ, & Risch NJ (2005). Genetic Structure, Self-Identified Race/Ethnicity, and Confounding in Case-Control Association Studies. Am J Hum Genet, 76, 268–275. [DOI] [PMC free article] [PubMed] [Google Scholar]
  66. Voracek M, & Loibl LM (2007). Genetics of suicide: a systematic review of twin studies. The Middle European Journal of Medicine, 119(15-16), 463–475. 10.1007/s00508-007-0823-2 [DOI] [PubMed] [Google Scholar]
  67. Wang Z, Yu C, Wang J, Bao J, Gao X, & Xiang H (2016). Age-period-cohort analysis of suicide mortality by gender among white and black Americans, 1983-2012. Int J Equity Health, 15(1), 107. 10.1186/s12939-016-0400-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  68. Wood PK, Steinley D, & Jackson KM (2015). Right-sizing statistical models for longitudinal data. Psychol Methods, 20(4), 470–488. 10.1037/met0000037 [DOI] [PMC free article] [PubMed] [Google Scholar]
  69. Wu Q, Zhang J, Walsh L, & Slesnick N (2022). Heterogeneous trajectories of suicidal ideation among homeless youth: predictors and suicide-related outcomes. Dev Psychopathol, 1–13. 10.1017/S0954579422000372 [DOI] [PMC free article] [PubMed] [Google Scholar]
  70. Xiao Y, Cerel J, & Mann JJ (2021). Temporal Trends in Suicidal Ideation and Attempts Among US Adolescents by Sex and Race/Ethnicity, 1991-2019. JAMA Netw Open, 4(6), e2113513. 10.1001/jamanetworkopen.2021.13513 [DOI] [PMC free article] [PubMed] [Google Scholar]
  71. Xiao Y, & Lindsey MA (2021). Racial/Ethnic, Sex, Sexual Orientation, and Socioeconomic Disparities in Suicidal Trajectories and Mental Health Treatment Among Adolescents Transitioning to Young Adulthood in the USA: A Population-Based Cohort Study. Adm Policy Ment Health, 48(5), 742–756. 10.1007/s10488-021-01122-w [DOI] [PMC free article] [PubMed] [Google Scholar]
  72. Zhu X, Tian L, & Huebner ES (2019). Trajectories of Suicidal Ideation from Middle Childhood to Early Adolescence: Risk and Protective Factors. J Youth Adolesc, 48(9), 1818–1834. 10.1007/s10964-019-01087-y [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplement
NIHMS2184429-supplement-Supplement.docx (27.9KB, docx)

RESOURCES