Preliminary findings on caffeine intake, screen time, social factors, and psychological well-being: their impact on chronotype and sleep health in Hispanic adolescents

Abstract

This study examined the relationships between caffeine intake, screen time, and chronotype/sleep outcomes in adolescents, with a focus on differences between Hispanic and non-Hispanic groups and the influence of peer network health, school environment, and psychological factors, including perceived stress, depression, and anxiety. Data from the Adolescent Brain Cognitive Development (ABCD) study were analyzed using t-tests and structural equation modeling (SEM) to assess behavioral, social, and psychological predictors of chronotype, social jet lag, and weekday sleep duration, incorporating demographic covariates. Hispanic adolescents exhibited a later chronotype (Cohen’s d = 0.42), greater social jet lag (Cohen’s d = 0.38), and shorter weekday sleep duration (Cohen’s d = −0.12) compared to non-Hispanic peers. They also reported higher caffeine intake (Cohen’s d = 0.22), though caffeine was not significantly associated with sleep outcomes. Screen time was more prevalent among Hispanic adolescents, particularly on weekday evenings (Cohen’s d = 0.27) and weekend evenings (Cohen’s d = 0.35), and was strongly associated with later chronotype and greater social jet lag. Higher perceived stress was linked to later chronotype and greater social jet lag, while depressive symptoms were associated with earlier chronotype and lower social jet lag. The SEM model explained 12.9% of variance in chronotype, 10.5% in social jet lag, and 6.2% in weekday sleep duration. These findings highlight disparities in adolescent sleep health but should be interpreted cautiously due to methodological limitations, including low caffeine use and assessment timing variability. Targeted interventions addressing screen time, peer relationships, and stress may improve sleep, while longitudinal research is needed to clarify causality.

Graphical Abstract

caffeine, stress, depression, anxiety, sleep hygiene, circadian rhythms, sleep deprivation, pediatrics—adolescents

Statement of Significance

Adolescence is a pivotal period for developing healthy sleep habits, yet many factors—such as screen time, caffeine use, social relationships, and psychological stress—can disrupt sleep. This study highlights significant disparities in sleep behaviors, caffeine intake, and psychological well-being between Hispanic and non-Hispanic adolescents, revealing how screen use, peer networks, depression, anxiety, and perceived stress influence sleep timing and quality. These findings underscore the need for culturally tailored interventions that address evening screen time, social influences, and stress regulation to improve adolescent sleep health. Future research should explore how modifying screen habits, fostering positive peer environments, and managing psychological stress can help mitigate long-term health risks and promote well-being in adolescents from diverse backgrounds.

Introduction

Adolescence is a period of profound biological, psychological, and social change, marked by significant shifts in sleep patterns. One critical dimension of sleep, chronotype—the individual preference for sleep timing—undergoes endogenous developmental changes during this stage, with a natural shift toward later sleep and wake times due to maturation of the circadian system [1]. However, this biological shift often conflicts with socially imposed schedules, such as early school start times, contributing to an overall reduction in sleep duration [2]. Chronotype plays a pivotal role in physical health [3], emotional regulation [4], and academic performance [5] during this developmental stage. Additionally, the modern adolescent lifestyle, characterized by increased screen time [6–8], caffeine consumption [9–12], and evolving social environments [13], exacerbates this misalignment by further delaying sleep onset. These disruptions lead to sleep health challenges such as social jet lag [14] and reduced sleep duration [15–17]. Both endogenous biological changes and external lifestyle factors contribute to adverse consequences, including impaired cognitive function [5, 18], increased risk of substance use [19], and poorer mental health [17, 20–22].

Hispanic adolescents face unique sociocultural contributors to chronotype misalignment and poor sleep health. Cultural norms, such as the frequent consumption of caffeinated beverages (e.g. coffee, energy drinks, and traditional drinks like café de olla), play a significant role. Compared to their non-Hispanic counterparts, Mexican-American adolescents and other Hispanic youth have higher rates of coffee consumption, earlier introduction to caffeinated beverages, and a strong cultural attachment to coffee-drinking practices [23, 24]. For example, approximately 74% of Hispanic Americans, including Mexican Americans, drink coffee daily—12 percentage points higher than other Americans [23]. Moreover, Mexican-American children often begin drinking coffee at a much younger age, as early as age five, compared to non-Hispanic youth, who typically start in adolescence [24]. These patterns reflect cultural traditions, such as enjoying coffee with baked goods, which hold emotional and social significance within Hispanic families [24]. Additionally, Hispanic consumers exhibit a preference for bold coffee flavors, emphasizing the deeply ingrained role of coffee in their sociocultural identity [24].

These unique patterns highlight the critical need to study the impact of caffeine consumption on Hispanic adolescents’ sleep health and chronotype. Frequent and early caffeine exposure in this population may exacerbate existing sleep disparities by contributing to delayed sleep timing, social jet lag, and shortened sleep duration. Despite the importance of these factors, caffeine consumption remains underexplored in studies of adolescent sleep health, particularly in the context of culturally specific behaviors and traditions.

Research indicates that environmental and behavioral factors are key determinants of adolescent sleep health. Screen time [8], particularly during evening hours [13], is associated with delayed sleep timing due to its impact on melatonin secretion and sleep latency. Similarly, caffeine intake [9–12], a common stimulant consumed by adolescents, can exacerbate sleep difficulties by prolonging sleep onset.

In parallel, the social and psychological context, including peer relationships, perceived stress, anxiety, depression, and school environment [25–27], shapes daily routines and sleep behaviors. For example, positive peer networks [28] may mitigate the effects of screen time [29] and caffeine intake [30–32] by promoting healthier sleep habits. Peer relationships become increasingly influential during adolescence, affecting bedtime routines, social activity timing, and overall sleep patterns. Supportive peer networks may encourage structured routines and discourage behaviors that disrupt sleep, such as late-night socialization or excessive caffeine use, while peer conflict or irregular peer schedules can contribute to delayed sleep timing and reduced sleep duration.

Beyond social influences, psychological factors, including stress, anxiety, and depression [26], also play a crucial role in adolescent sleep health. Perceived stress is associated with disrupted sleep [27], as elevated stress levels can lead to hyperarousal, difficulty initiating sleep, and increased social jet lag. Adolescents experiencing higher stress, anxiety, or depressive symptoms often exhibit shorter sleep duration, poorer sleep quality, and a later chronotype [33] due to dysregulated circadian rhythms and maladaptive coping behaviors, such as increased evening screen use or caffeine consumption.

Finally, the school environment profoundly shapes sleep behaviors [25] by imposing daily schedules, influencing stress levels, and determining opportunities for social interaction and extracurricular engagement. Supportive school settings, characterized by positive relationships and low stress, can promote better sleep hygiene, while unsupportive environments may exacerbate irregular sleep patterns or reduce sleep duration. Together, these social and contextual factors represent key moderators in understanding the complex interplay between behavioral, psychological, and environmental influences on adolescent sleep health, providing crucial insights for targeted interventions to optimize chronotype/sleep outcomes.

Despite the growing recognition of these influences, there remains a need for integrated studies that explore the interplay between behavioral, social, and psychological factors in shaping adolescent Hispanic chronotype and sleep health. The current study addresses this gap by leveraging data from the Adolescent Brain Cognitive Development (ABCD) study, a nationally representative cohort that provides a unique opportunity to investigate these relationships.

This study aims to examine the associations between caffeine intake, screen time, and chronotype/sleep outcomes in Hispanic adolescents, while exploring the moderating effects of peer network health, perceived stress, anxiety, depression, and school environment. It is hypothesized that higher caffeine intake and increased screen time will be associated with delayed chronotype and poorer sleep health, while peer network health may buffer these adverse effects. Additionally, perceived stress, anxiety, and depressive symptoms are expected to negatively impact sleep outcomes, while a supportive school environment may mitigate these effects. By identifying key modifiable factors, this research seeks to inform culturally tailored interventions that promote healthier sleep behaviors during adolescence, a critical period for establishing lifelong patterns of well-being.

Methods

Study design and participants

This study utilized data from the ABCD study [34], a longitudinal, nationally representative cohort of adolescents in the United States. The current analysis focused on data from baseline through year 4 (see Supplementary Figure 1 showing years of data collection for structural equation modeling [SEM] across variables), including ages 10–12 years (see Table 1). Participants with complete data on key variables, including chronotype, caffeine intake, screen time, and moderating factors, were included in the analysis. The final analytic sample comprised 2788 adolescents.

Table 1.

Key variables in SEM

Variable	Mean ± SD	%
Age (years)	11.11 ± 1.00
Sex (female)		46.80
Hispanic ethnicity		17.20
Normalized household income	1.25 ± 1.08
Caffeine Intake 1	0.13 ± 0.69
Caffeine Intake 3	0.03 ± 0.18
Caffeine Intake 4	0.44 ± 1.83
Caffeine Intake 6	1.09 ± 3.60
Caffeine Intake 9	0.02 ± 0.36
Caffeine maximum	12.95 ± 16.02
Chronotype (HH:MM)	28:18 ± 1:26
Social jet lag (h)	2.20 ± 0.26
Weekly sleep duration (h)	8.56 ± 1.24
School environment	20.53 ± 2.49
Peer network health	11.68 ± 7.84
Depression	3.99 ± 0.09
Anxiety	3.99 ± 0.09
Perceived stress	1.73 ± 0.32

Descriptive statistics for key variables included in the SEM. Mean and standard deviation (SD) are reported for continuous variables, including age, normalized household income, caffeine intake measures, chronotype, social jet lag, weekly sleep duration, school environment, peer network health, depression, anxiety, and perceived stress. Percentages are provided for categorical variables, such as sex (female) and Hispanic ethnicity. Chronotype is reported in HH:MM format, while social jet lag and weekly sleep duration are reported in hours. These variables represent the behavioral, social, and demographic factors used to investigate chronotype/sleep outcomes in adolescents.

Measures

The years of data collection for different variables were chosen based on the data availability in the ABCD dataset and model temporality (predictor < moderator < outcome variables). Caffeine intake was only available at baseline. Screen time, school environment, peer network health had the largest amount of data available at baseline, year 1, and year 2, respectively. Depression and anxiety were temporally chosen from year 2, in addition to perceived stress from year 3, based on data availability as potential moderators. Chronotype/sleep outcome measures were available at years 2 and 4, but utilized from year 4 to avoid any temporal overlap with year 2 variables.

Household income

Household income data at baseline were derived from the ABCD Longitudinal Parent Demographics Survey [35]. Income levels were categorized on a 10-point scale, with each point corresponding to a specific income range. The scale ranged from 1 to 10, where 1 represented an annual income of less than $5000, and 10 indicated incomes of $200 000 or more. The intermediate categories were as follows: 2 ($5000–$11 999), 3 ($12 000–$15 999), 4 ($16 000–$24 999), 5 ($25 000–$34 999), 6 ($35 000–$49 999), 7 ($50 000–$74 999), 8 ($75 000–$99 999), and 9 ($100 000–$199 999). Household income was adjusted for the number of household members to account for differences in household size.

Hispanic classification

Participants self-reported their racial and ethnic backgrounds at baseline, which were used to classify them into distinct groups. For the purposes of this study, race–ethnicity was recoded into a binary variable: Hispanic participants were assigned a value of 1, while all other racial–ethnic groups were assigned a value of 0.

Chronotype and sleep outcomes

Chronotype was evaluated using year 4 data from the ABCD study, collected via the Munich Chronotype Questionnaire (MCTQ) [36]. Chronotype was determined based on the midpoint of sleep. The MCTQ also provided a framework for measuring social jet lag [2], which quantifies the misalignment between an individual’s internal circadian rhythm and their social schedule. To enhance precision and better accommodate variability in chronotypes, a sleep-corrected measure [2] of social jet lag was calculated by comparing sleep onset times on free days versus workdays. This adjustment offers a more accurate representation of circadian misalignment in diverse populations. In addition to chronotype and social jet lag, average weekly sleep duration from the MCTQ was included to assess overall sleep patterns.

Behavioral factors

Caffeine intake at baseline

Caffeine consumption was measured using self-reported items on the frequency and quantity of caffeine intake over the past 6 months [37]. This included specific indicators such as weekly consumption of coffee (e.g. brewed or flavored types); tea (e.g. green, black, sweet, or Earl Grey); caffeinated soda (e.g. Mountain Dew, Coke, Pepsi, Dr. Pepper); and energy drinks (e.g. Red Bull, Monster, 5-Hour Energy).

We included every available measure of caffeine intake from the dataset when building the initial latent factor model for caffeine. Maximum caffeine intake was retained in the final model because it had a high factor loading, indicating that it strongly contributed to the latent caffeine construct. Although maximum caffeine intake may represent a single extreme instance, it could serve as a proxy for habitual high caffeine consumption patterns or individual caffeine sensitivity, both of which may have lasting effects on adolescent sleep and chronotype. Adolescents’ caffeine intake tends to remain relatively stable over time [38], particularly for high consumers, making earlier measures relevant for later sleep-related outcomes. Additionally, episodic high caffeine intake may reflect behavioral patterns (e.g. using caffeine to counteract sleep deprivation), reinforcing long-term sleep disruptions that could influence chronotype.

The SEM model incorporates a select subset of caffeine intake questions rather than the full set of prompts to ensure the latent variable accurately represents the core aspects of caffeine consumption patterns relevant to the study objectives. Questions were selected based on their psychometric properties, including strong factor loadings and conceptual alignment with the study’s focus on habitual and maximum intake behaviors. This approach balances model parsimony with theoretical rigor, avoiding redundancy while maintaining robust measurement of the latent construct. The question prompts associated with the variables included were:

Caffeine Intake 1: “How many servings of coffee (instant or brewed, including flavored types) with caffeine do you consume? Enter the amount in cups (8 oz) as a decimal.”

Caffeine Intake 3: “How many servings of espresso-based drinks (e.g. Latte, Mocha, Americano) with caffeine do you consume? Enter the amount in shots of espresso as a decimal.”

Caffeine Intake 4: “How many servings of tea with caffeine (e.g. Green Tea, Black Tea, Sweet Tea, Earl Grey) do you consume? Enter the amount in cups (8 oz) as a decimal.”

Caffeine Intake 6: “How many servings of soda with caffeine (e.g. Mountain Dew, Jolt, Coke, Pepsi, Dr. Pepper, Barq’s Root Beer) do you consume? Enter the amount in cans (12 oz) as a decimal.”

Caffeine Intake 9: “How many servings of energy drinks (e.g. Red Bull, Monster, Rock Star, 5-Hour Energy, AMP, Full Throttle) do you consume? Enter the amount in typical serving sizes as a decimal.”

Additionally, participants reported the largest amount of caffeine consumed in a single day (Caffeine Maximum) and the type of beverage consumed during that day. A latent variable for caffeine intake was constructed to encapsulate overall patterns of caffeine consumption, including frequency, quantity, and peak intake.

Screen time at baseline

Screen time was assessed using self-reported measures of weekday and weekend usage across various activities. Weekday screen time included hours spent watching TV or movies, watching videos such as YouTube, and playing video games on devices like computers or consoles. Similar measures captured weekend usage, such as hours spent on these activities during weekends. Additional measures included time spent texting and engaging with social media. A comprehensive latent variable was constructed to summarize patterns of total screen time use [39].

Moderators

Anxiety and depression at year 2

Parent-reported depression and anxiety scores at year 2 were assessed using the Child Behavior Checklist (CBCL), specifically the DSM-5 Depression and Anxiety T-scores, which reflect parent-reported depressive or anxiety symptom severity in 6- to 18-year-olds. These measures capture clinical levels of depressive or anxiety symptoms, with higher scores indicating greater symptom severity, and scores greater than 65 suggesting pathology [40, 41].

Perceived stress at year 3

Parent-reported perceived stress at year 3 was assessed using the 10-item Perceived Stress Scale (PSS-10), a widely used measure evaluating the frequency of stress-related thoughts and feelings over the past month. The PSS-10 captures subjective perceptions of stress, including the extent to which individuals feel overwhelmed, out of control, or unable to cope with daily demands. Higher scores indicate greater perceived stress, with established cutoffs suggesting that elevated scores are associated with increased risk for stress-related health outcomes. This measure has been validated across diverse populations and is commonly used in research to assess psychological stress in both clinical and non-clinical settings [42, 43].

Peer network health at year 2

The Peer Network Health scale [44] was used at year 2 to evaluate the protective qualities of participants’ peer networks. This scale generates a composite score by summing responses regarding each peer’s substance use, behavioral influence, and activity types. Participants reported on risky or negative behaviors, including whether they knew each nominated peer used substances, whether the peer was a daily user, and whether they were influenced by the peer to use or avoid substances. They also identified engagement in illegal, violent, or dangerous activities. Additionally, participants described positive or protective interactions with their peers, such as receiving help with schoolwork or transportation and providing support through problem discussions. These responses are combined into a total score for each peer, calculated using a weighted system that ranges from −14 to 14. Higher scores reflect healthier peer networks, while lower scores indicate higher behavioral risk. The scale demonstrates strong internal reliability (Cronbach’s α = 0.84) and significant correlations with self-reported measures of substance use, including alcohol and marijuana use [44].

School environment at year 1

The School Environment subscale, derived from the School Risk and Protective Factors (SRPF) [45] survey, comprises six items designed to evaluate students’ connection to their school environment. This subscale assesses key dimensions, including the availability of extracurricular activities, relationships with teachers, receipt of praise for good performance, opportunities to engage in class activities, communication between the school and parents, and students’ sense of safety at school. We utilized year 1 data for our analyses.

Covariates

Covariates included demographic variables such as Hispanic ethnicity classification, age, sex, and total household income normalized by number of household members and family identifier to account for clustering within families.

Statistical analysis

Exploratory analyses

Independent samples t-tests

To examine differences between Hispanic and non-Hispanic adolescents across sleep-related, caffeine intake, screen time, and social/environmental measures, we conducted independent samples t-tests. Each t-test assessed the mean differences between groups for continuous variables, assuming unequal variances where appropriate based on Levene’s test for homogeneity of variance. Effect sizes were calculated using Cohen’s d, with small (0.2), moderate (0.5), and large (0.8) thresholds used for interpretation [46].

Multiple comparisons correction

Given the number of t-tests conducted across interrelated variables (e.g. multiple caffeine intake measures, screen time across different time periods, and sleep variables), we applied the Benjamini–Hochberg (BH) procedure to control the false discovery rate (FDR) [47]. This approach balances Type I error control while maintaining statistical power, adjusting p-values to reduce the likelihood of false-positive findings due to multiple testing. Adjusted p-values below .05 were considered statistically significant.

Hypothesis-based analyses

SEM

A SEM was employed to assess the relationships between caffeine intake, screen time, and chronotype/sleep outcomes. The model examined direct effects of caffeine intake and race/ethnicity on chronotype/sleep measures and included additional predictors such as peer network protection, depression/anxiety, perceived stress, and school environment. Interaction terms were used to explore potential moderation effects. Covariates (age, sex, household income, and screen time) were included to control for confounders (Figure 1).

Figure 1.

SEM path diagram. SEM diagram illustrating the temporal relationships from left to right between demographic covariates (e.g. income), screen time variables, peer network health, depression, anxiety, perceived stress, and chronotype/sleep outcomes (chronotype, social jet lag, and weekday sleep duration). Significant paths (p < .05) are emphasized; caffeine intake has been omitted due to lack of significance. This model highlights the interplay between behavioral, social, and demographic factors in predicting adolescent chronotype/sleep outcomes.

The latent variable for caffeine intake was constructed using four observed indicators, and residual covariances among indicators were specified based on theoretical assumptions. Chronotype/sleep outcomes (chronotype, social jet lag, and weekly sleep duration) were allowed to correlate. Model fit was evaluated using several indices, including the chi-square test, comparative fit index (CFI), Tucker–Lewis index (TLI), root mean square error of approximation (RMSEA), and standardized root mean square residual (SRMR). The final model was assessed against established criteria for acceptable fit: CFI and TLI values greater than 0.90, RMSEA below 0.08, and SRMR at or below 0.05 [48, 49]. Data analyses were performed using R version 4.2.2 [50], RStudio Version 2023.09.1+494 [51], and the lavaan package [52], with a significance threshold set at p < .05 for all statistical tests. Effect sizes were calculated using established thresholds of 0.2 for small, 0.5 for medium, and 0.8 for large effects, following standard guidelines [46].

Moderation analysis

Interaction terms were included to assess moderation effects of peer network health, depression/anxiety symptoms, perceived stress, and school environment on the relationship between caffeine intake, screen time, and chronotype/sleep outcomes.

Simplified SEM for race-stratified analyses

A simplified SEM was constructed separately for each racial-ethnic subgroup to examine the relationships between caffeine consumption, socioeconomic factors, depression, and social jet lag. This model was designed in response to the challenges encountered when attempting to estimate a more comprehensive model incorporating additional behavioral and environmental influences, which failed to converge when stratified by race. The simplified model retained key predictors while reducing complexity to improve model stability and comparability across groups.

In this model, social jet lag was predicted by caffeine consumption, depressive symptoms, and socioeconomic status, allowing for an assessment of the direct relationships among these factors. Additionally, caffeine consumption was modeled as an outcome of depressive symptoms and socioeconomic status to evaluate whether these factors indirectly influenced social jet lag through caffeine use. However, because the simplified model included fewer predictors than the full model, it provided only a limited explanation of variability in social jet lag.

Approach to addressing outliers and skewness

In the present study, we carefully evaluated and addressed issues of outliers and skewness in the dataset to ensure the robustness of our analyses. Prior to conducting the primary statistical analyses, all variables were examined for outliers using standardized residuals, z-scores, and visual inspection of boxplots. Observations with z-scores exceeding ±3.29 were flagged as potential outliers, consistent with recommendations for large datasets. Extreme values were subsequently cross-checked against the original data to confirm their validity, and any data entry errors were corrected. Valid extreme observations were retained unless they exerted undue influence, assessed via Cook’s distance and leverage diagnostics, in which case sensitivity analyses were performed to determine their impact on results [53].

Skewness and kurtosis were evaluated for all continuous variables using descriptive statistics and histograms. Variables with skewness or kurtosis values exceeding ±2 were considered for transformation, as these thresholds indicate substantial departure from normality that could bias parameter estimates, particularly in maximum likelihood estimation methods. For highly skewed variables, log or square root transformations were applied to reduce asymmetry. When transformations failed to achieve normality or interpretability was compromised, robust statistical techniques that accommodate non-normal data, such as bootstrapping and robust standard errors, were employed.

Latent variables were constructed using observed indicators, and the assumption of multivariate normality was assessed for these constructs. Multivariate outliers were identified using Mahalanobis distance, with a chi-square threshold (p < .001) used to flag cases for further investigation [54]. Influential multivariate outliers were examined and, where necessary, excluded from the final analysis to ensure that parameter estimates were not disproportionately influenced.

Ethical considerations

The ABCD study received ethical approval from the University of California San Diego IRB, and all participants provided informed consent/assent.

Results

Exploratory analyses (Table 2)

Table 2.

Comparison of chronotype/sleep, caffeine intake, screen time, and social factors between Hispanic and non-Hispanic adolescents

Variable	Non-Hispanic Mean (SD)	Hispanic Mean (SD)	t-Statistic	Adjusted p-value	Effect size (Cohen’s d)
Chronotype (HH:MM)	28:16 (1:21)	28:43 (1:24)	8.16	1.00E-05	0.42
Social jet lag	2.19 (0.25)	2.28 (0.27)	7.21	3.00E-05	0.38
Weekday sleep duration (h)	8.60 (1.15)	8.46 (1.23)	−2.26	0.021	−0.12
Caffeine Intake 1	1.03 (0.12)	1.06 (0.18)	3.97	0.0002	0.25
Caffeine Intake 3	1.01 (0.04)	1.02 (0.08)	3.67	0.0011	0.26
Caffeine Intake 4	1.12 (0.34)	1.08 (0.29)	−2.06	0.048	−0.09
Caffeine Intake 6	1.27 (0.50)	1.38 (0.56)	4.03	0.0002	0.22
Caffeine Intake 9	1.00 (0.00)	1.00 (0.00)	−0.56	0.578	−0.02
Caffeine maximum	3.13 (1.81)	3.52 (1.70)	4.52	0.0001	0.22
Weekday screen time 1	1.38 (0.33)	1.42 (0.31)	2.68	0.008	0.13
Weekday screen time 2	1.27 (0.33)	1.36 (0.36)	5.18	1.00E-06	0.27
Weekday screen time 3	1.30 (0.32)	1.34 (0.34)	2.24	0.022	0.12
Weekday screen time 4	1.06 (0.12)	1.08 (0.14)	3.15	0.002	0.18
Weekday screen time 5	1.02 (0.07)	1.02 (0.08)	0.69	0.52	0.04
Weekend screen time 7	1.52 (1.20)	1.65 (1.23)	2.02	0.048	0.1
Weekend screen time 8	1.33 (0.38)	1.47 (0.41)	6.54	1.00E-07	0.35
Weekend screen time 9	1.41 (0.38)	1.46 (0.42)	1.93	0.058	0.1
Weekend screen time 10	1.07 (0.16)	1.10 (0.20)	2.89	0.003	0.17
Weekend screen time 12	1.06 (0.16)	1.09 (0.19)	2.76	0.006	0.16
Screen time total (weekday)	1.05 (0.11)	1.07 (0.14)	2.86	0.003	0.17
Screen time total (weekend)	1.16 (0.27)	1.20 (0.29)	2.85	0.005	0.15
School environment	20.54 (2.47)	20.50 (2.58)	−0.35	0.73	−0.02
Peer network health	11.51 (7.81)	12.50 (7.96)	2.51	0.012	0.13
Depression	3.99 (0.09)	3.99 (0.09)	−0.07	0.944	−0.003
Anxiety	3.99 (0.09)	3.99 (0.09)	0.91	0.325	0.05
Perceived stress	1.75 (0.30)	1.66 (0.37)	−4.82	2.00E-06	−0.28

Comparison of sleep, caffeine intake, screen time, and social factors between Hispanic and non-Hispanic adolescents. Mean values and standard deviations (SD) are provided for both groups across various variables. Statistical comparisons are summarized using t-statistics, adjusted p-values (corrected using the Benjamini–Hochberg procedure to control for false discovery rate), and effect sizes (Cohen’s d). Significant differences (adjusted p < .05) were observed across multiple measures, highlighting key behavioral and environmental disparities. Hispanic adolescents exhibited later chronotypes, greater social jet lag, and higher caffeine and screen time measures compared to their non-Hispanic peers. Additionally, Hispanic adolescents reported lower perceived stress levels, while differences in depression and anxiety scores were not statistically significant.

Chronotype and sleep outcomes

Hispanic adolescents exhibited a later chronotype (mean sleep midpoint: 28:43 HH:MM) compared to non-Hispanic peers (28:16 HH:MM; p < .0001, Cohen’s d = 0.42), reflecting a moderate effect size. Similarly, social jet lag was significantly greater among Hispanic adolescents (2.28 hours) than non-Hispanic adolescents (2.19 hours; p < .0001, Cohen’s d = 0.38), also indicating a moderate effect. Weekday sleep duration was slightly shorter for Hispanic adolescents (8.46 hours) than for non-Hispanics (8.60 hours; p = .021, Cohen’s d = −0.12), though this effect size was small.

Caffeine intake

Hispanic adolescents reported higher caffeine intake across multiple measures. Caffeine Intake 1 was significantly higher (1.06 vs. 1.03; p < .0002, Cohen’s d = 0.25), as was Caffeine Intake 3 (1.02 vs. 1.01; p = .0011, Cohen’s d = 0.26), both showing small-to-moderate effect sizes. Caffeine Intake 6 was also significantly higher in Hispanic adolescents (1.38 vs. 1.27; p < .0002, Cohen’s d = 0.22). Additionally, maximum caffeine intake was higher for Hispanic adolescents (3.52 vs. 3.13; p < .0001, Cohen’s d = 0.22), though the effect size remained small.

Screen time

Significant differences emerged in screen time patterns between groups. Hispanic adolescents reported more screen time on weekday afternoons (1.42 vs. 1.38; p = .008, Cohen’s d = 0.13) and weekday evenings (1.36 vs. 1.27; p < .0001, Cohen’s d = 0.27), with a small effect for afternoons and a moderate effect for evenings. On weekends, Hispanic adolescents also engaged in more evening screen time (1.47 vs. 1.33; p < .0001, Cohen’s d = 0.35), demonstrating a moderate effect size.

Social and environmental factors

Hispanic adolescents reported higher peer network health scores than non-Hispanic adolescents (12.50 vs. 11.51; p = .012, Cohen’s d = 0.13), reflecting a small effect and suggesting stronger perceived social support. Perceived stress was significantly lower among Hispanic adolescents (1.66 vs. 1.75; p < .0001, Cohen’s d = −0.28), with a moderate effect size. In contrast, depression (p = .944, Cohen’s d = −0.003) and anxiety (p = .325, Cohen’s d = 0.05) scores did not significantly differ between groups.

Hypothesis-based analyses

SEM fit

The SEM demonstrated excellent fit to the data, as indicated by a CFI of 0.966 (robust CFI = 0.970) and a TLI of 0.930 (robust TLI = 0.939). The RMSEA was 0.025 (90% CI: 0.021–0.028; robust RMSEA = 0.023, 90% CI: 0.019–0.027), and the SRMR was 0.013. Collectively, these indices suggest that the hypothesized model provides a strong representation of the observed data (Table 3).

Table 3.

SEM results for chronotype/sleep outcomes and predictors

Outcome	Predictor	Estimate	Std. Error	z-value	P	Std. Estimate	Significance
Chronotype	Hispanic ethnicity	0.069	0.027	2.579	.010	0.043	*
	Caffeine intake (latent)	−0.016	0.026	−0.587	.557	−0.013	NS
	Age	0.040	0.023	1.702	.089	0.029	NS
	Parental income	0.104	0.026	4.048	<.001	0.076	**
	Peer network health	0.086	0.025	3.501	<.001	0.063	**
	Depression (CBCL)	−0.078	0.024	−3.283	.001	−0.061	**
	Anxiety (CBCL)	−0.044	0.030	−1.454	.146	−0.032	NS
	Perceived stress (PSS)	0.086	0.025	3.501	<.001	0.063	**
	Screen time (weekday evenings)	0.137	0.045	3.035	.002	0.101	**
	Screen time (weekend evenings)	0.107	0.044	2.395	.017	0.078	*
Social jet lag	Hispanic ethnicity	0.069	0.027	2.579	.010	0.043	*
	Caffeine intake (latent)	−0.016	0.026	−0.587	.557	−0.013	NS
	Age	−0.004	0.005	−0.859	.390	−0.015	NS
	Parental income	0.012	0.005	2.302	.021	0.048	*
	Peer network health	0.017	0.005	3.684	<.001	0.068	**
	Depression (CBCL)	−0.014	0.005	−3.029	.002	−0.058	**
	Anxiety (CBCL)	−0.005	0.006	−0.881	.379	−0.019	NS
	Perceived stress (PSS)	0.017	0.005	3.684	<.001	0.068	**
	Screen time (weekday afternoons)	0.014	0.007	2.039	.041	0.056	*
	Screen time (weekend evenings)	0.018	0.009	2.118	.034	0.072	*
Weekday sleep duration	Hispanic ethnicity	0.069	0.027	2.579	.010	0.009	*
	Caffeine intake (latent)	−0.016	0.026	−0.587	.557	−0.003	NS
	Age	−0.068	0.021	−3.261	.001	−0.058	**
	Parental income	−0.004	0.025	−0.159	.873	−0.003	NS
	Peer network health	−0.023	0.022	−1.044	.296	−0.020	NS
	Depression (CBCL)	0.062	0.026	2.381	.017	0.053	*
	Anxiety (CBCL)	−0.005	0.006	−0.881	.379	−0.019	NS
	Perceived stress (PSS)	−0.023	0.022	−1.044	.296	−0.020	NS
	Screen time (weekday afternoons)	−0.093	0.041	−2.254	.024	−0.079	*
	Screen time (weekend evenings)	−0.076	0.033	−2.294	.022	−0.065	*

Results from the SEM examining predictors of sleep outcomes (chronotype, social jet lag, and weekday sleep duration). Predictors include Hispanic ethnicity, caffeine intake (latent variable), age, parental income, peer network health, depression (CBCL), anxiety (CBCL), perceived stress (PSS), and screen time (weekday and weekend).

Key findings include significant positive associations between Hispanic ethnicity and all three sleep outcomes, as well as screen time (weekday and weekend evenings) with chronotype and social jet lag. Depression was negatively associated with chronotype and social jet lag but positively associated with weekday sleep duration, while anxiety did not show significant relationships. Perceived stress was positively associated with chronotype and social jet lag but not with sleep duration.

Caffeine intake did not emerge as a significant predictor of any sleep outcome. These results highlight the varying roles of psychological, behavioral, and socioeconomic factors in shaping adolescent sleep patterns. The mixed findings for screen time suggest that its impact varies depending on timing and context, with evening screen use being more strongly associated with later chronotype and greater social jet lag, while afternoon screen use was linked to reduced weekday sleep duration.

Mixed findings for screen time—(1) Chronotype: weekday evening screen time was positively associated (Std. Estimate = 0.101, p = .002); weekend evening screen time was positively associated (Std. Estimate = 0.078, p = .017). (2) Social jet lag: weekday afternoon screen time was positively associated (Std. Estimate = 0.056, p = .041); weekend evening screen time was positively associated (Std. Estimate = 0.072, p = .034). (3) Weekday sleep duration: weekday afternoon screen time was negatively associated (Std. Estimate = −0.079, p = .024); weekend evening screen time was negatively associated (Std. Estimate = −0.065, p = .022).

Significance levels: **p < .01, *p < .05, NS: not significant.

Latent variable for caffeine intake

The latent variable for caffeine intake was defined by four indicators (Caffeine Intake 1, Caffeine Intake 4, Caffeine Intake 6, Caffeine Max). All indicators showed significant standardized loadings, ranging from 0.213 to 0.767, demonstrating their substantial contributions to the latent construct.

Predictors of Chronotype and Sleep Outcomes

Chronotype

Hispanic ethnicity was associated with later sleep midpoints (b = 0.069, p = .010), reflecting ethnic differences in sleep timing. Peer network health positively predicted later sleep midpoints (b = 0.086, p < .001), underscoring the influence of social dynamics on sleep patterns. Higher household income was significantly associated with later sleep midpoints (b = 0.104, p < .001), suggesting socioeconomic factors impact sleep timing. Screen time during weekday evenings (b = 0.137, p = .002) and weekend evenings (b = 0.107, p = .017) contributed to later chronotype.

Depression scores were negatively associated with later sleep midpoints (b = −0.078, p = .001), suggesting that higher depressive symptoms were linked to earlier sleep timing. Anxiety was not significantly associated with chronotype (b = −0.044, p = .146). Perceived stress was positively associated with later sleep midpoints (b = 0.086, p < .001), indicating that higher stress levels were linked to delayed sleep timing.

In contrast, caffeine intake was not a significant predictor of chronotype (b = −0.016, p = .557).

Social jet lag

Hispanic ethnicity was associated with greater social jet lag (b = 0.069, p = .010), as was peer network health (b = 0.017, p < .001), reinforcing the role of ethnic and social influences. Weekend evening screen time significantly predicted greater social jet lag (b = 0.018, p = .034), and weekday afternoon screen time was also a predictor (b = 0.014, p = .041).

Depression was negatively associated with social jet lag (b = −0.014, p = .002), suggesting that higher depressive symptoms were linked to smaller weekday–weekend sleep timing discrepancies. Anxiety was not significantly associated with social jet lag (b = −0.005, p = .379). Perceived stress was positively associated with greater social jet lag (b = 0.017, p < .001), implying higher stress levels were linked to more pronounced discrepancies in sleep timing across weekdays and weekends.

Caffeine intake was not a significant predictor of social jet lag (b = −0.016, p = .557).

Weekday sleep duration

Age was a significant predictor of shorter weekday sleep duration (b = −0.068, p = .001), indicating that older adolescents experienced reduced sleep duration on school nights. Screen time during weekday afternoons negatively impacted sleep duration (b = −0.093, p = .024), while weekend evening screen time also showed a negative association (b = −0.076, p = .022).

Depression was positively associated with weekday sleep duration (b = 0.062, p = .017), suggesting that higher depressive symptoms were linked to longer weekday sleep duration. Anxiety did not show a significant association with weekday sleep duration (b = −0.005, p = .379). Perceived stress was not significantly associated with weekday sleep duration (b = −0.023, p = .296).

Peer network health, household income, and caffeine intake did not exhibit significant relationships with weekday sleep duration.

Mediation effects of caffeine intake

Caffeine intake did not mediate the relationships between Hispanic ethnicity or other predictors and chronotype/sleep outcomes. For instance, its effect on the midpoint of sleep on free days was non-significant (b = −0.016, p = .557).

Variance explained

The model explained 12.9% of the variance in sleep midpoint on free days, 10.5% in social jet lag, and 6.2% in weekday sleep duration. The latent variable for caffeine intake accounted for 19.5% of the variance in caffeine consumption. These values suggest moderate explanatory power of the predictors included in the model.

Simplified race-stratified analyses

In comparison to the full model, which incorporated a broader set of factors such as environmental influences, sleep behaviors, and social determinants, the simplified model produced stronger model fit statistics for some racial subgroups but explained very little variance in social jet lag. Other race-stratified models exhibited poor fit, indicating that the reduced model structure did not adequately capture key predictors across all groups. In contrast, the more comprehensive model, though designed to account for a wider range of interacting variables, failed to converge when stratified by race, highlighting the complexity of modeling social jet lag within distinct racial-ethnic populations. These findings suggest that while caffeine intake and parental income may have small effects, additional behavioral, environmental, and sociocultural factors are needed to better explain variability in social jet lag (Supplementary Tables 1–4).

Discussion

The present study investigated the intricate relationships between caffeine intake, screen time, and chronotype/sleep outcomes in adolescents, with a focus on moderating factors such as Hispanic ethnicity, peer network health, depression, anxiety, and perceived stress using data from the ABCD study. Through t-test comparisons, distinct behavioral and social differences emerged between Hispanic and non-Hispanic adolescents, revealing differences in chronotype, social jet lag, weekday sleep duration, caffeine consumption, screen time behaviors, and stress-related factors. These group differences provided a critical foundation for further exploration through SEM, which illuminated the multifaceted interplay of behavioral, social, and environmental factors impacting adolescent sleep health.

Hispanic adolescents exhibited a later chronotype, greater social jet lag, and shorter weekday sleep duration compared to their non-Hispanic peers. These findings align with previous research emphasizing the role of cultural and socio-environmental contexts [55–60] in shaping sleep patterns. Cultural norms, family routines, and social obligations [55, 57–59] may contribute to later sleep timing in Hispanic youth, reinforcing the need for further investigation into these mechanisms in diverse populations.

The observed higher caffeine intake among Hispanic adolescents across multiple metrics supports findings from studies indicating greater consumption of caffeinated beverages in certain minority populations [23, 24]. This behavior may be influenced by socioeconomic and cultural factors, such as targeted marketing, family consumption patterns, or the use of caffeine to counteract sleep deprivation. Despite higher intake levels, caffeine was not significantly associated with chronotype or social jet lag in the SEM results, suggesting that its role in shifting sleep timing may be less pronounced than other behavioral and environmental influences.

Screen time patterns also differed significantly between groups, with Hispanic adolescents reporting more screen use on weekday afternoons and evenings and during weekends. These findings align with prior studies indicating that minority adolescents may engage in more recreational screen use (e.g. watching videos or using social media) as a coping mechanism [61]. Evening screen time was associated with delayed chronotype and increased social jet lag, supporting existing evidence that blue light exposure and the stimulating nature of screen activities disrupt circadian rhythms and sleep patterns [62].

Social and emotional factors also played a significant role. Hispanic adolescents reported higher peer network health scores compared to non-Hispanics, consistent with prior research highlighting collectivist cultural norms and strong social support structures in Hispanic communities [63]. Peer network health was significantly associated with later chronotype and greater social jet lag, suggesting that strong peer relationships might encourage behaviors such as late-night socialization, electronic communication, or co-sleeping arrangements that delay sleep timing. Perceived stress, in contrast, was significantly lower among Hispanic adolescents, potentially reflecting cultural differences in stress appraisal or coping mechanisms. Lower perceived stress was associated with earlier chronotype and lower social jet lag, reinforcing prior research linking high stress levels to disrupted sleep timing [27].

Lastly, the modest but significant effect sizes observed in this study, particularly for caffeine intake and screen time, underscore the multifaceted nature of sleep health disparities. While these behaviors contribute to group differences, they represent only part of a broader constellation of factors, including environmental, cultural, and socioeconomic influences, that shape adolescent sleep health.

For the SEM, Hispanic ethnicity emerged as a consistent predictor across all three chronotype/sleep outcomes: chronotype, social jet lag, and weekday sleep duration. These findings align with the t-test differences noted above.

The observed relationship between higher depressive symptoms and earlier sleep timing and reduced social jet lag is consistent with prior literature linking depressive symptoms to phase-advanced sleep patterns in adolescents [64]. Research suggests that adolescents with elevated depressive symptoms often exhibit earlier bedtimes and wake times, which may reflect increased fatigue, diminished social engagement, or altered homeostatic sleep regulation. This contrasts with findings in adult populations, where depression is frequently associated with eveningness and delayed sleep timing [65]. The sleep outcome results align with the hypersomnia subtype of depression, which is characterized by excessive sleepiness and prolonged sleep duration [66]. The positive association between depressive symptoms and weekday sleep duration suggests that adolescents experiencing greater depressive symptoms may compensate for increased fatigue with extended sleep on school nights. However, the reduced social jet lag among those with greater depressive symptoms may indicate less engagement in weekend social activities that would otherwise shift sleep timing, a pattern consistent with social withdrawal behaviors commonly associated with depression [67].

Unlike depression, anxiety did not show significant associations with chronotype, social jet lag, or weekday sleep duration, suggesting that its impact on adolescent sleep timing may be more subtle. Prior studies have linked anxiety to increased sleep latency, frequent night-time awakenings, and overall poorer sleep quality [68], rather than shifts in sleep timing or duration. The lack of significant associations in the present study could be due to the measurement approach, as chronotype and social jet lag primarily reflect sleep timing rather than fragmentation or quality. Adolescents with elevated anxiety may experience greater physiological hyperarousal, which could impair sleep onset and maintenance without necessarily altering their circadian preference or weekday–weekend sleep timing discrepancies. Another possibility is that the effects of anxiety on sleep timing may be masked by co-occurring depression, as depressive symptoms more strongly predicted chronotype and social jet lag in this analysis.

Higher perceived stress was linked to later sleep timing and greater social jet lag, a pattern consistent with prior findings that stress disrupts sleep–wake regulation and promotes circadian misalignment [27]. Adolescents experiencing chronic stress may engage in maladaptive coping behaviors, such as increased evening screen use or late-night socialization, which delay sleep timing and contribute to greater variability in weekday–weekend sleep schedules. The association between higher stress levels and greater social jet lag suggests that stressed adolescents may attempt to compensate for weekday sleep deficits by shifting their sleep schedules on weekends, a behavior that may reinforce further circadian misalignment. Interestingly, perceived stress did not significantly predict weekday sleep duration, which contrasts with some studies linking chronic stress to shorter sleep duration due to heightened physiological arousal [69]. One explanation may be that adolescents facing high stress compensate with longer weekend sleep rather than altering weekday sleep. The findings highlight the bidirectional relationship between stress and sleep, where poor sleep patterns can exacerbate stress responses, further disrupting sleep stability.

Peer network health was significantly associated with later chronotype and greater social jet lag, indicating that supportive peer environments may act as a buffer against behavioral risks that impact sleep patterns. This underscores the importance of social connectedness in adolescent well-being, particularly as young people navigate increased autonomy and screen use. Promoting positive peer dynamics could play a key role in improving chronotype outcomes for adolescents.

The findings related to screen time revealed that its impact on chronotype/sleep outcomes depends on both timing and context. Weekday screen time, particularly in the afternoons, was associated with reduced weekday sleep duration, suggesting that excessive or poorly timed device use may encroach on sleep. Additionally, weekend evening screen time was positively associated with increased social jet lag and reduced weekday sleep duration, highlighting how late-night use disrupts sleep–wake consistency. Conversely, evening screen time was positively associated with later chronotype, reflecting its role in delaying bedtime. These results support the need for tailored screen time interventions that emphasize timing and duration, particularly targeting evenings and weekend use.

Socioeconomic status, represented by household income, significantly predicted chronotype/sleep outcomes, reflecting the broader impact of socioeconomic factors on adolescent health behaviors and light exposure. Adolescents from higher-income households may experience greater access to resources that influence light exposure patterns, such as access to outdoor spaces or structured routines that align with natural light cycles, whereas those from lower-income households may face environmental factors like increased exposure to artificial light or irregular schedules that disrupt circadian rhythms. Neighborhood environments and stressors associated with financial insecurity may also shape light exposure and, consequently, chronotype/sleep outcomes.

Although caffeine intake was hypothesized to play a significant role, it was not strongly associated with chronotype/sleep outcomes in this sample. This suggests that its effects on adolescent sleep may be more subtle. The MCTQ primarily measures circadian factors, and there is limited evidence that caffeine directly impacts the circadian system in adolescents. Instead, caffeine consumption may be more closely linked to sleep quality rather than to fundamental shifts in sleep timing or chronotype. Factors such as the timing of consumption or individual sensitivity to caffeine, which were not fully captured in this study, may moderate the relationship. Future research should explore temporal patterns of caffeine use, its potential effects on sleep architecture and quality, and possible moderating variables such as stress or sleep hygiene practices.

These findings have several implications for interventions aimed at improving adolescent sleep health. Interventions should be culturally tailored to address the unique sleep challenges faced by adolescents from different racial and ethnic backgrounds. Strategies that leverage positive peer relationships could also enhance chronotype/sleep outcomes, particularly if peers model healthy sleep behaviors. Additionally, encouraging families to establish screen-free routines, especially during evenings and weekends, could mitigate the disruptive effects of screen use on sleep. Finally, a holistic approach that incorporates socioeconomic factors and addresses systemic inequities is essential to reduce disparities in adolescent sleep health.

This study has limitations. The cross-sectional design precludes causal inferences, and longitudinal analyses are needed to better understand temporal dynamics. The reliance on self-reported measures for some variables may introduce reporting bias, and unmeasured confounders, such as dietary patterns or psychological stress, could influence the observed relationships. Additionally, the time elapsed between assessments may have allowed for other influencing factors to modify the observed correlations. The timing of key measures in this study varied, with caffeine intake assessed at baseline, peer network health at year 2, and sleep outcomes at year 4. This inconsistency limits the ability to establish direct causal links, as adolescent behaviors, such as caffeine consumption and peer influence, fluctuate over time. For example, the correlation between screen time and sleep outcomes may reflect cumulative behavioral reinforcement, wherein prolonged exposure to screen-based activities leads to habitual changes in sleep patterns over time. Increased exposure to artificial light from screens may also alter circadian rhythms through delayed melatonin onset [70, 71]. In contrast, the lack of correlation with caffeine intake could stem from variability in individual tolerance, metabolism, and adaptive responses to habitual consumption [72]. Another major limitation of the study is the overall low caffeine consumption reported by participants, which may have reduced the ability to detect meaningful associations. Additionally, relying solely on baseline caffeine data ignores potential increases in use as adolescents age, limiting the ability to capture long-term effects on sleep outcomes. Furthermore, caffeine effects on sleep may be more immediate or transient, making them harder to capture in cross-sectional analyses. A key analytical limitation was the inability to conduct a race-stratified SEM due to convergence issues. The ABCD dataset combines race and ethnicity into a single categorical variable, and when attempting a race-stratified analysis, the small sample sizes of some racial groups—particularly Asian (n = 90) and Black (n = 246) participants—relative to the large number of estimated parameters (540) led to model instability and non-convergence. Given these challenges and the study’s primary focus on Hispanic ethnicity rather than racial differences, we proceeded with the model in its current form to provide the clearest and most interpretable findings. The difficulty in constructing even a simplified race-stratified SEM suggests that the findings may have limited generalizability across racial groups. Future models that incorporate a broader range of sleep-related behaviors, social influences, and contextual factors may provide a more comprehensive understanding of the mechanisms contributing to racial-ethnic disparities in social jet lag. Despite these limitations, the findings underscore the multifaceted influences on adolescent chronotype/sleep outcomes and highlight the need for culturally informed, context-specific strategies to address sleep health. As screen use and caffeine consumption continue to rise, understanding how these behaviors interact with individual and environmental factors will be critical for improving adolescent Hispanic health outcomes.

These findings highlight the nuanced interplay of behavioral, social, and environmental factors in shaping adolescent Hispanic sleep health. Interventions should prioritize managing screen time, particularly during evenings and weekends, and fostering supportive peer relationships. Additionally, the role of perceived stress in delaying sleep timing and exacerbating social jet lag underscores the need for stress-reduction strategies, such as mindfulness-based interventions or structured routines, to promote healthier sleep patterns. Addressing the higher caffeine intake and screen time observed in Hispanic adolescents could be critical to improving their chronotype/sleep outcomes. Future research should explore longitudinal trajectories and include diverse populations to enhance generalizability. A holistic approach targeting screen use, stress regulation, and social influences may better support adolescent well-being and reduce the long-term health consequences of disrupted sleep.

Acknowledgments

We would like to express our deepest gratitude to the participants and their families for their invaluable contributions to the Adolescent Brain Cognitive Development (ABCD) study. Your dedication to participating in this longitudinal research has provided critical insights into adolescent development and has advanced our understanding of behavioral, social, and environmental factors influencing health. We also extend our appreciation to the researchers, staff, and collaborators involved in the ABCD study for their commitment to collecting, curating, and sharing this rich dataset. Your efforts make studies like ours possible and drive scientific progress. This work would not have been possible without your generosity and commitment to improving adolescent health and well-being for future generations. Thank you.

Funding

This research was supported by the National Institute on Alcohol Abuse and Alcoholism through grant K23 AA026869 awarded to A.D.M. and the Department of Veterans Affairs through Merit Award BX003431 granted to M.J.M.

Disclosure statement

Financial disclosure: none.

Non-financial disclosure: none.

Author contributions

Alexander L. Wallace (Conceptualization [supporting], Methodology [supporting], Writing—review & editing [supporting]), Laika Aguinaldo (Resources [supporting], Writing—review & editing [supporting]), Michael L. Thomas (Formal analysis [supporting], Methodology [supporting], Visualization [supporting], Writing—review & editing [supporting]), Michael J. McCarthy (Funding acquisition [supporting], Writing—review & editing [supporting]), and Alejandro D. Meruelo (Conceptualization [lead], Formal analysis [lead], Funding acquisition [lead], Investigation [lead], Methodology [lead], Visualization [lead], Writing—original draft [lead], Writing—review & editing [lead]).

Data availability

This study utilized data from the Adolescent Brain Cognitive Development (ABCD) study (https://abcdstudy.org), funded by the National Institutes of Health (NIH) under the following grant numbers: U01DA041022, U01DA041028, U01DA041048, U01DA041089, U01DA041106, U01DA041117, U01DA041120, U01DA041134, U01DA041148, U01DA041156, U01DA041174, U24DA041123, and U24DA041147. The data analyzed are publicly available through the National Institute of Mental Health Data Archive (NDA) as part of the ABCD Study.