Available via license: CC BY 4.0
Content may be subject to copyright.
Integrating multi-system environmental factors to predict brain and behavior in
adolescents
RH: Predicting brain and behavior via multi-system factors
Jivesh Ramduny, Ph.D., Samuel Paskewitz, Ph.D., Inti A. Brazil, Ph.D., Arielle Baskin-Sommers,
Ph.D.
Drs. Ramduny, Paskewitz, and Baskin-Sommers are with the Department of Psychology, Yale
University. Dr. Ramduny also is with the Kavli Institute for Neuroscience, Yale University. Dr.
Baskin-Sommers also is with the Department of Psychiatry and Wu Tsai Institute, Yale University.
Dr. Brazil is with the Donders Institute for Brain, Cognition and Behaviour, Radboud University. Dr.
Brazil also is with the Forensic Psychiatric Centre Pompestichting, The Netherlands.
Samuel Paskewitz, Ph.D., served as the statistical expert for this research.
Acknowledgments
Data used in the preparation of this article were obtained from the ABCD Study® (abcdstudy.org/),
held in the NIMH Data Archive (NDA). This is a multisite, longitudinal study designed to recruit
more than 10,000 children aged 9-10 and follow them over 10 years into early adulthood. The ABCD
Study is supported by the National Institutes of Health (NIH) and additional federal partners under
award numbers: U01DA041048, U01DA050989, U01DA051016, U01DA041022, U01DA051018,
U01DA051037, U01DA050987, U01DA041174, U01DA041106, U01DA041117, U01DA041028,
U01DA041134, U01DA050988, U01DA051039, U01DA041156, U01DA041025, U01DA041120,
U01DA051038, U01DA041148, U01DA041093, U01DA041089, U24DA041123, U24DA041147.
The full list of federal supporters is available at https://abcdstudy.org/federal-partners.html. The
complete lists of participating sites and study investigators can be found at
https://abcdstudy.org/consortium_members/. The ABCD Consortium investigators designed and
implemented the study and/or provided the data but did not necessarily participate in the analysis or
writing of this report. This manuscript reflects the views of the authors and may not reflect the
opinions or views of the NIH or ABCD Consortium investigators. Additional support for this work
was made possible from NIEHS R01-ES032295, R01-ES031074, and R21DA057592. This work also
obtained support from the Yale Kavli Institute for Neuroscience and the Wu Tsai Institute at Yale
University. We thank the Yale Center for Research Computing for guidance and use of the research
computing infrastructure.
Disclosure
Drs. Ramduny, Paskewitz, Brazil, and Baskin-Sommers have reported no biomedical financial
interests or potential conflicts of interest.
Correspondence
Jivesh Ramduny, Ph.D., 100 College St, New Haven, CT 06520-8047, USA; Phone: (+1) 203-432-
5759; email: jivesh.ramduny@yale.edu.
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
2
Abstract
Objective: Environmental factors have long been shown to influence brain structure and
adolescent psychopathology. However, almost no research has included environmental
factors spanning micro-to-macro-systems, brain structure, and psychopathology in an
integrated framework. Here, we assessed the ways and degree to which multi-system
environmental factors during late childhood predict subcortical volume and psychopathology
during early adolescence.
Method: We used the baseline and 2-year follow-up data from the Adolescent Brain
Cognitive DevelopmentSM Study (N = 2,766). A Bayesian latent profile analysis was applied
to obtain distinct multi-system environmental profiles during late childhood. The profiles
were used in a path analysis to predict their direct and indirect effects on subcortical volume
and psychopathology during early adolescence.
Results: Bayesian latent profile analysis revealed nine environmental profiles. Two distinct
profiles predicted greater externalizing problems in adolescents: (i) adversity across, family,
school, and neighborhood systems and (ii) family conflict and low school involvement. In
contrast, a profile of family and neighborhood affluence predicted fewer externalizing
difficulties. Further, family and neighborhood affluence predicted higher subcortical volume,
which in turn, predicted fewer externalizing problems; whereas, family economic and
neighborhood adversity predicted lower subcortical volume, which in turn, predicted greater
externalizing difficulties.
Conclusion: We captured direct and indirect influences of environmental factors across
multiple systems on externalizing psychopathology. Specifying the equifinal pathways to
externalizing psychopathology serves to provide an evidence base for establishing different
types of interventions based on the needs and risk profiles of youth.
Diversity and Inclusion Statement
The current study is part of the ongoing Adolescent Brain Cognitive DevelopmentSM Study
(ABCD Study®) for which youth are recruited from elementary schools in the United States
that are informed by gender, race, ethnicity, socioeconomic status, and urbanicity. The
ABCD Study® aims to recruit youth longitudinally by sampling the sociodemographic
makeup of the US population. Two of the authors self-identifies as a member of one or more
historically underrepresented racial and/or ethnic groups in science. One of the authors
identifies as a part of an underrepresented gender group in science. The authors also are
representative of the communities for which data was collected and contributed to design,
analysis, and/or interpretation of the work. Finally, every effort was made to cite the work of
authors from underrepresented and minoritized groups in academic research.
Keywords: adolescence, environment, subcortical brain volume, psychopathology,
externalizing
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
3
Introduction
An extensive body of research has identified that adolescent psychopathology relates to
changes in brain development and experiences in the environment. However, much of this
work has been siloed into work specifying the environmental or neurobiological factors
related to psychopathology adolescence.
On the one hand, decades of research have documented that environmental systems influence
the development of psychopathology (i.e., externalizing and internalizing problems)1-5. Meta-
analyses report medium-to-large effects between adversity in adolescents’ families (e.g.,
conflict, caregiver nonacceptance) and neighborhoods (e.g., experiencing violence or
disadvantage) and adolescent psychopathology6,7. Externalizing problems have been
associated with adverse experiences in the form of family poverty, harsh parenting,
association with deviant peers, concentrated disadvantage, and exposure to community
violence2. Similarly, internalizing problems have been linked to maternal depression,
maltreatment by a caregiver, peer victimization, and exposure to community violence can
exacerbate internalizing difficulties in youth3-5. On the other hand, neurobiological theories of
adolescent psychopathology have emphasized that structural changes, a general indicator of
brain health. In particular, subcortical brain volumes, are sensitive to externalizing and
internalizing problems during adolescent development8,9. These brain regions support self-
regulation and affective processing10, and differences in their structural volumes have been
related to various aspects of adolescent psychopathology11.
Integrating environmental and neurobiological factors, some researchers have shown that
experiences within different environmental systems influence structural brain development12.
Adverse experiences—such as low household income, harsh parenting, maltreatment, peer
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
4
victimization, neighborhood disadvantage, and community violence—have shown to be
associated with reduced subcortical volumes in regions that are involved in cognitive and
emotion processing13-16. For example, children from lower income families and those
exposed to community violence have smaller hippocampal and amygdala volumes13-16.
Additionally, maternal harsh parenting has been related to smaller amygdala volume17 and
childhood maltreatment has been associated with smaller hippocampal volume18. Researchers
also have shown that anti-poverty policies in the United States mitigate socioeconomic-
related brain differences, such that in high cost of living states with generous anti-poverty
programs, the association between family income and hippocampal volume resembled that of
lower cost of living states14. However, the majority of these studies have focused on
examining the influences of single environmental systems on brain structures, and have not
examined the combined influence of family, school, neighborhood, and policy factors.
Moreover, many of these studies have been designed with relatively small sample sizes, more
restricted sociodemographic samples (e.g., all low income, all high income, all maltreatment,
all disadvantaged neighborhood), and a focus on selective subcortical regions (e.g.,
hippocampus and amygdala).
Research on the influence of environmental experiences on adolescent psychopathology,
brain on psychopathology, and their interactions (albeit in limited ways) has laid a strong
foundation for understanding how psychopathology may unfold in different contexts for
different youth. However, there is a need to employ an integrated approach19,20 that fully
captures the transactions among multiple environmental systems, brain development, and
psychopathology. First, we must do better in estimating the multiple, and often interacting,
environmental systems youth encounter—some that relate to a youth’s immediate
surroundings (e.g., family, school, neighborhood)21 while others that relate to larger societal
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
5
and cultural contexts (e.g., policies, law)22. Second, we need to understand how
interconnected environmental experiences, directly or indirectly, relate to subcortical brain
structure and psychopathology.
In the present study, we tested the relationships among multiple environmental systems,
subcortical gray matter (GM) volume, and psychopathology using the Adolescent Brain
Cognitive DevelopmentSM Study (ABCD Study®)23. First, we aimed to identify distinct
profiles of youth during late childhood from family, school, neighborhood, and policy
systems using a novel Bayesian latent profile analysis (LPA)24 method. Most often,
researchers use multiple regression and factor analysis with single, or possibly two,
environmental systems and subcortical ROIs, limiting their ability to capture complex
(within-person) interactions of multi-system environmental factors and their associations with
brain structures. LPA is an analytic approach that derive unique profiles of individuals that
exhibit similar characteristics (e.g., environmental experiences). Here, we used the Bayesian
LPA method as it has been shown to capture more nuanced and more certain profiles than
conventional LPA24. We hypothesized the emergence of latent profiles characterized by
moderate and high adversities across multiple environmental systems. Given the novelty of
combining multi-system environments in the Bayesian LPA method, we did not have specific
hypotheses about latent profiles that could capture more subtle variations in adversities.
Second, we aimed to assess the ways and degree to which the multi-system environmental
profiles during late childhood predict subcortical GM volume and psychopathology (i.e.,
externalizing, internalizing) during early adolescence. Based on prior research1-5,13-16, we
hypothesized that profiles describing high adverse family and neighborhood environments
would predict smaller subcortical GM volume and greater externalizing problems. Following
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
6
evidence that brain structures can mediate the association between poverty and externalizing
problems during adolescence16,25, we also expected an indirect relationship between adverse
multi-system environments and externalizing via subcortical GM volume.
Methods
Participants
The ABCD Study is a ten-year longitudinal study that tracks the development of children and
adolescents across 21 research sites in the United States23. We used the demographic,
environmental, behavioral, and imaging data from the ABCD Study Release 5.0 (N=11,868)
at baseline (9-10 years) and 2-year follow-up (11-12 years) (https://abcdstudy.org/). The
ABCD Study obtained approval from a centralized Institutional Review Board (IRB) located
at the University of California, San Diego in addition to obtaining local IRB approval from
each of the imaging sites. Written assent was provided by the youth and written informed
consent was obtained by their parents or guardians. Only participants who had complete
demographic, environmental, psychopathology, and imaging data in addition to passing
successfully the MRI data quality control criteria described by the ABCD Data Analysis and
Informatics Center (DAIC)26 (https://wiki.abcdstudy.org/release-notes/imaging/quality-
control.html) were included in the study. A total of 2,766 youth remained in the study (Table
1).
Environmental Data
The environmental factors were obtained at baseline from the ABCD Study Culture and
Environment27, Linked External Data28, and Adolescent Neural Urbanome29 batteries. We
focused on multi-system environmental factors that have shown relationships with subcortical
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
7
GM volume and youth psychopathology1-5,13-16. Choosing environmental factors across
multiple systems from the ABCD Study is challenging as they show some degree of
correlation with each other. Further, the Bayesian LPA method that we used to identify
distinct profiles of youth requires that the factors follow a normal distribution24. Therefore,
environmental factors that had binary outcomes were not included in this study. Policy data
available in the ABCD Study, such as medicaid expansion, naloxone policies, and Good
Samaritan law were binary outcomes and had no variability in their distributions28. By
contrast, the marijuana laws variable was not binary, and it displayed sufficient variability
within the ABCD sample.
Family Conflict
The Family Conflict subscale from the Family Environment Scale (FES) consists of nine
items, indicating whether each statement is True or False by the youth for most family
members. The nine items were summed and reverse scored to obtain a measure of family
conflict for each youth. A higher value on the FES Conflict subscale indicates less family
conflict perceived by the youth.
Parenting Style
The Acceptance Scale from the Children’s Reports of Parental Behavior Inventory (CRPBI)
consists of five items, where the youth describe their caregivers’ parenting style on a 3-point
scale. The mean score from the five items was used as a measure of parenting style for each
youth. A higher value on the CRPBI Acceptance Scale indicates warmer parenting style
perceived by the youth.
Income-to-needs Ratio
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
8
The income-to-needs ratio was calculated as the median of the income band described by the
ABCD Study divided by the federal poverty level based on the respective household size.
The median of the first income band was set at $5,000 and the median for the last income
band was set at $200,000. The federal poverty level was obtained from the Department of
Health and Human Services (https://www.healthcare.gov/glossary/federal-poverty-level-fpl/).
An income-to-needs ratio of 1 indicates living at the poverty threshold, a ratio greater than 1
denotes living above the poverty threshold, and a ratio less than 1 denotes living below the
poverty threshold.
School Involvement
The School Involvement subscale from the School Risk and Protective Factors (SRPF)
questionnaire contains four items as indicators of positive involvement in school. The scores
from the four items were summed to obtain a measure of school involvement for each youth.
A higher value on the SRPF School Involvement subscale indicates more school involvement
perceived by the youth.
Neighborhood Deprivation
The area deprivation index (ADI) reflects the weighted sum of 17 composite scores related to
employment, education, income and poverty, and housing using the youth’s home address
from the 2011-2015 American Community Survey (Table S1, available online). We reverse
coded the ADI scores with higher ADI values indicating lower neighborhood disadvantage.
Neighborhood Safety and Crime
The Safety from Crime item scales from the PhenX Toolkit describe three statements
administered to the parent related to feeling safe walking in their neighborhood, violence in
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
9
their neighborhood, and crime in their neighborhood. Each item is rated on a 5-point Likert
scale ranging from “strongly agree (5)” to “strongly disagree (1)”. Only one statement was
administered to youth to assess their feelings about safety and crime in their neighborhood.
The scores from the parent and youth item scales were summed to obtain a measure of
neighborhood safety and crime for each youth. A higher value on the Safety from Crime item
scales indicates a safer neighborhood perceived by the parent and youth.
Residential Segregation
The index of concentration at the extremes (ICE) was used to examine the extent to which a
population in a specified area is concentrated into the wealthiest and poorest extremes of a
specified social distributions. ICE measures the distributions of affluence and poverty within
racial or ethnic groups across the wealthiest and poorest areas of the community based on the
2014-2018 American Community Survey. It ranges from -1 to 1 such that a positive value
indicates concentration of a racial/ethnic group in affluent areas, a negative value denotes
concentration of a racial/ethnic group in impoverished areas, and a value closer to 0 indicates
no concentration of a racial/ethnic group in either affluent or poorer areas of the community.
Marijuana Laws
Currently, there are 38 states that have legalized cannabis for medical use, 24 states that
provide legal access to cannabis for recreational use, and 9 states that allow low THC, high
CBD products either for medical purposes or as a legal defense in the United States
(http://www.ncsl.org/research/health/state-medical-marijuana-laws.aspx). States that legalize
either recreational or medical cannabis use reflect more liberal marijuana laws as opposed to
those which forbid legal access to cannabis, therefore being more conservative. We obtained
policy data representing marijuana laws across the United States, and reverse coded the
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
10
categories for cannabis legalization. Cannabis legalization was categorized as follows: 1:No
legal access to cannabis; 2:Low THC/CBD; 3:Medical; and 4:Recreational. A higher value in
the cannabis legalization categories indicates a less conservative law.
Psychopathology Data
The Child Behavior Checklist (CBCL) is a parent-report assessment that was used to measure
externalizing and internalizing behaviors30. We used the parent-report as youth self-report on
externalizing and internalizing were not available for the 2-year follow-up. Externalizing
behaviors tend to reflect symptoms such as aggression and rule-breaking, whereas
internalizing symptoms capture anxiety and depression. The CBCL externalizing and
internalizing scores were obtained from their respective syndrome scales that were
subsequently 𝑻-standardized. The higher the CBCL externalizing and internalizing 𝑻-scores,
the greater the risk of experiencing behavioral and emotional problems.
ABCD Study
N
%
Mean (SD)
Range
Sample Size
2,766
100
—-
—-
Race
White
1,685
60.9
—-
—-
Black
247
8.9
—-
—-
Asian
68
2.5
—-
—-
AIANa
30
1.1
—-
—-
NHPIb
5
0.2
—-
—-
Mixed
57
2.1
—-
—-
Otherc
57
2.1
—-
—-
Unspecified
208
7.5
—-
—-
Ethnicity
Hispanic or Latino
409
14.8
—-
—-
Sexd
Male
1,380
49.9
—-
—-
Female
1,386
50.1
—-
—-
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
11
Educatione
< HS Diploma
105
3.8
—
—
HS Diploma/GED
232
8.4
—
—
Some College
486
17.6
—
—
Associate Degree
387
14.0
—
—
Bachelor Degree
907
32.8
—
—
Post Graduate Degree
649
23.5
—
—
Family Incomef
< $5,000
54
2.0
—-
—-
$5,000-$11,999
72
2.6
—-
—-
$12,000-$15,999
61
2.2
—-
—-
$16,000-$24,999
120
4.3
—-
—-
$25,000-$34,999
186
6.7
—-
—-
$35,000-$49,999
276
10.0
—-
—-
$50,000-$74,999
427
15.4
—-
—-
$75,000-$99,999
538
19.5
—-
—-
$100,000-$199,999
1,032
37.3
—-
—-
> $200,000
0
0
—-
—-
Environmental Factors
Family Conflictg
2,766
—-
2.45 (1.95)
0-9
Parenting Styleg
2,766
—-
2.79 (0.29)
1-3
Income-to-Needs Ratiog
2,766
—-
3.21 (2.27)
0.12-9.92
School Involvementh
2,766
—-
13.22 (2.22)
5-16
Neighborhood Deprivationi
2,766
—-
96.78 (16.21)
3.73-124.52
Neighborhood Safety & Crimei
2,766
—-
15.87 (3.09)
4-20
Residential Segregationi
2,766
—-
0.15 (0.27)
-0.73-0.86
Marijuana Lawsj
2,766
—-
2.29 (0.73)
1-4
Psychopathology
CBCL-Externalizingk
2,766
—-
44.19 (9.59)
33-83
CBCL-Internalizingk
2,766
—-
48.03 (10.42)
33-82
Table 1. Demographic, environmental, and behavioral characteristics derived from the ABCD
Study NIMH Data Archive Release 5.0. aAIAN = American Indian and Alaska Native. bNHPI =
Native Hawaiian and Pacific Islander. cOther race/ethnicity corresponds to Eastern and Western
European, Afro-Carribean/Indo-Carribbean/West Indian, Middle Eastern/North African in addition to
parents who selected “Other race” to indicate that the predefined groups did not apply to them.
dParticipant sex denotes youth’s sex assigned at birth. eEducation refers to the highest grade or level of
school a parent has completed or the highest degree they have received. fFamily Income refers to the
total income in a household and the income bands are provided by the ABCD Study. gFamily-related
environmental factors describe family conflict, parenting style, and income-to-needs ratio at baseline.
hSchool-related environmental factor describes school involvement at baseline. iNeighborhood-related
environmental factors describe neighborhood deprivation, neighborhood crime and safety, and
residential segregation at baseline. jPolicy-related environmental factor refers to marijuana laws with
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
12
regards to cannabis legalization in the United States at baseline. kCBCL = Child Behavior Checklist
indexing externalizing and internalizing psychopathology at 2-year follow-up. Note that the CBCL
externalizing and internalizing scores were 𝑻-standardized. We also compared the characteristics of
our sample with the full ABCD sample to test for differences in demographics, multi-system
environments, and psychopathology (Supplement 1).
MRI Data Acquisition
Structural T1-weighted and T2-weighted MRI scans were acquired using Siemens Prisma,
Philips, and GE 750 3T scanners with a 32-channel head coil23. 3D MPRAGE T1-weighted
and 3D FSE T2-weighted volumes with spatial resolution 1x1x1mm3 were obtained for each
youth at 2-year follow-up. The structural MRI data were preprocessed using the DAIC
standard processing pipeline26.
Subcortical Gray Matter Volume
For each participant, the subcortical GM structures were labeled using an automated, atlas-
based, volumetric segmentation procedure from FreeSurfer26. These structures include 19
ROIs corresponding to bilateral nucleus accumbens, amygdala, caudate, hippocampus,
pallidum, putamen, thalamus, ventral diencephalon, cerebellum, and midline brainstem.
Bayesian Latent Profile Analysis
We performed a Bayesian latent profile analysis (LPA) using the environmental factors. LPA
explains a set of indicator variables by grouping participants into latent profiles, i.e.,
categories of individuals with similar characteristics. The LPA model assumes that each
participant belongs to a single latent profile and that indicator variables have independent
normal likelihoods with means that vary across profiles24. Conventional methods for
determining the correct number of latent profiles often give conflicting results for which
number of profiles is best to select. We used a Bayesian form of LPA based on the Dirichlet
process mixture model62 that automatically detects the correct number of latent profiles. This
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
13
Bayesian LPA method was implemented using the Python package vbayesfa24. We used
proportional reduction in classification entropy to assess the distinctiveness of the profiles
inferred.
Integrated Approach
We examined relationships among multi-system environmental profiles, subcortical GM
volume, and externalizing and internalizing using a path analysis (Figure 1). The path
analysis was conducted by including only one sibling, i.e., one child per family, to limit
family dependency confounds, with the lavaan package in R version 4.3.2. We tested the
direct and indirect effects of the multi-system environmental profiles simultaneously—a
direct effect represents a relationship between a multi-system environmental profile and
subcortical GM volume or between a multi-system environmental profile and
externalizing/internalizing whereas an indirect effect represents a relationship between a
multi-system environmental profile and externalizing/internalizing via the subcortical GM
volume. For direct effects, the coefficient estimates, standardized errors, and statistical
significance were reported. For indirect effects, the coefficient estimates and 95% confidence
intervals (CIs), which were obtained from a bootstrapping procedure, were reported.
Figure 1. Integrated Approach. 1. The ABCD Study® NDA Release 5.0 was used to obtain
environmental, brain, and behavioral data from late childhood (baseline) to early adolescence (2-year
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
14
follow-up). 2. The environmental factors were obtained at baseline and they captured multiple
systems including family, school, neighborhood, and policy. Family-related factors correspond to (A)
family conflict, (B) parenting style, and (C) income-to-needs ratio. School-related factor corresponds
to (D) school involvement. Neighborhood-related factors correspond to (E) neighborhood
disadvantage, (F) neighborhood safety and crime, and (G) residential segregation. Policy-related
factor corresponds to (H) marijuana laws with regards to cannabis legalization in the United States. 3.
The environmental factors were then used to generate distinct profiles (denoted by P1, P2, …., Pn) of
youth automatically that share similar characteristics without a priori specifying the number of latent
profiles using a Bayesian latent profile analysis framework. We assigned each participant to their
most probable profile and then treated the profiles like an observed variable. 4. Subsequently, the
subcortical gray matter (GM) volume was obtained at 2-year follow-up from the ABCD Study. The
subcortical GM volume corresponds to 19 ROIs including the bilateral nucleus accumbens, amygdala,
caudate, hippocampus, pallidum, putamen, thalamus, ventral diencephalon, cerebellum, and extending
to the midline brainstem. The behavioral measures also were obtained from the Child Behavior
Checklist (CBCL) which correspond to externalizing and internalizing psychopathology at 2-year
follow-up. 5. The integrated approach represents a path analysis linking the multi-system
environmental factors (denoted by P1, P2, …., Pn), subcortical GM volume, and externalizing and
internalizing psychopathology. A direct effect captures the relationship between a multi-system
environmental profile and subcortical GM volume or between a multi-system environmental profile
and externalizing/internalizing behavior. An indirect direct effect captures the relationship between a
multi-system environmental profile and externalizing/internalizing behavior via the subcortical GM
volume.
Results
Description of the multi-system environmental profiles
The Bayesian LPA model produced 9 distinct profiles with an excellent proportional
reduction in entropy of 0.90 (Figure 2). In creating labels, we wanted to use judgment-free
language and avoid listing factors as positive or negative. Further, given the array of factors
representing environmental systems, it was difficult to come up with a labeling scheme that
would apply appropriately to all factors. Therefore, we opted to use descriptive labels for
each profile based on the collection of factors that deviated from the mean. For each profile,
the characteristics of all environmental factors are provided in Table 2.
Profile 1 was characterized by below average family income [mean income band=6]
representing 37% of the sample. Profile 2 was characterized by family [mean income
band=9] and neighborhood affluence [ICE
%
%
%
%
%
=0.424; PhenX
%
%
%
%
%
%
%
%
=17.6; ADI
%
%
%
%
%
=85.5] representing 16%
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
15
of the sample. Profile 3 was characterized by family economic [mean income band=5] and
neighborhood adversity [PhenX
%
%
%
%
%
%
%
%
=12.2; ICE
%
%
%
%
%
=-0.246; ADI
%
%
%
%
%
=113] representing 15% of the
sample. Profile 4 was characterized by family economic affluence [mean income band=9]
representing 14% of the sample. Profile 5 was characterized by adversity across family
[CRPBI
%
%
%
%
%
%
%
%
=2.05; FES
%
%
%
%
%
=4.00; mean income band=5], school [SRPF
%
%
%
%
%
%
%
=11.0], and neighborhood
systems [PhenX
%
%
%
%
%
%
%
%
=14.1; ICE
%
%
%
%
%
=-0.000004; ADI
%
%
%
%
%
=106] representing 5% of the sample. Profile 6
was characterized by neighborhood affluence [ICE
%
%
%
%
%
=0.441; ADI
%
%
%
%
%
=70.3] and liberal marijuana
policy [mȷ%
%
%
%
=1.74], but below average family income [mean income band=6] representing 4%
of the sample. Profile 7 was characterized by adverse family interactions [CRPBI
%
%
%
%
%
%
%
%
=2.21;
FES
%
%
%
%
%
=3.12] and low school involvement [SRPF
%
%
%
%
%
%
%
=11.4], but family economic [mean income
band=8], and neighborhood affluence [ICE
%
%
%
%
%
=0.309] representing 4% of the sample. Profile 8
was characterized by neighborhood [ICE
%
%
%
%
%
=0.534; ADI
%
%
%
%
%
=40.8] and family economic affluence
[mean income band = 8], with somewhat liberal marijuana policy [mȷ%
%
%
%
=1.91] representing 3%
of the sample. Profile 9 was characterized by family conflict [FES
%
%
%
%
%
=5.56] and low school
involvement [SRPF
%
%
%
%
%
%
%
=8.79] representing 2% of the sample. The multi-system environmental
profiles differed significantly in the distributions of participant race/ethnicity, US region, and
sex assigned at birth (Figure S1, available online).
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
16
Figure 2. Descriptions of the multi-system environmental profiles obtained from family, school,
neighborhood, and policy factors. Family factors correspond to family conflict, parenting style, and
income-to-needs ratio. The school factor corresponds to school involvement. Neighborhood factors
correspond to neighborhood disadvantage, neighborhood safety and crime, and residential
segregation. The policy factor corresponds to marijuana laws. The proportion of youth in each distinct
profile also is displayed. Note that family conflict, neighborhood disadvantage, and marijuana laws
have been reverse coded. Dotted gray line denotes responses 0.50±SD from the mean of each profile.
Solid gray line denotes responses 1.0±SD from the mean of each profile.
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
17
Profile
Mean
Household
Size
Mean
Income
Band
ABCD
Income Range
Family
Conflict
(FES)
Parenting
Style
(CRPBI)
School
Involvement
(SRPF)
Neighborhood
Disadvantage
(ADI)
Neighborhood
Safety&Crime
(PhenX)
Residential
Segregation
(ICE)
Marijuana
Laws
(mj)
1
5
6
$35,000-$49,000
1.74 (0-9)
2.86 (2.2-3.0)
13.6 (6-16)
99.9 (72.2-116)
16.4 (6-20)
0.155 (-0.39-0.63)
2.38 (1-4)
2
5
9
$100,000-$199,999
1.38 (0-8)
2.89 (2.4-3.0)
13.6 (6-16)
85.5 (60.0-102)
17.6 (7-20)
0.424 (0.11-0.74)
2.13 (1-3)
3
5
5
$25,000-$34,999
2.10 (0-9)
2.86 (2.2-3.0)
13.9 (7-16)
113 (87.4-125)
12.2 (4-20)
-0.246 (-0.73-0.09)
2.44 (1-4)
4
4
9
$100,000-$199,999
1.69 (0-8)
2.86 (2.4-3.0)
13.1 (6-16)
101 (69.8-119)
16.3 (5-20)
0.095 (-0.44-0.41)
2.29 (1-4)
5
5
5
$25,000-$34,999
4.00 (0-9)
2.05 (1.0-2.4)
11.0 (4-16)
106 (68.0-125)
14.1 (5-20)
-0.000004 (-0.54-0.53)
2.37 (1-4)
6
5
6
$35,000-$49,999
1.88 (0-8)
2.86 (2.2-3.0)
12.8 (8-16)
70.3 (40.8-90.0)
16.6 (7-20)
0.441 (-0.21-0.74)
1.74 (1-3)
7
5
8
$75,000-$99,999
3.12 (0-9)
2.21 (1.0-2.6)
11.4 (4-16)
91.4 (55.3-115)
16.6 (11-20)
0.309 (-0.13-0.69)
2.15 (1-3)
8
4
8
$75,000-$99,999
1.59 (0-8)
2.77 (1.2-3.0)
12.9 (8-16)
40.8 (3.73-61.7)
17.0 (12-20)
0.534 (0.08-0.86)
1.91 (1-3)
9
6
6
$35,000-$49,999
5.56 (2-9)
2.69 (2.4-3.0)
8.79 (5-13)
99.8 (83.6-116)
16.3 (6-20)
0.183 (-0.31-0.54)
2.57 (1-3)
Table 2. Characteristics of the multi-system environmental profiles. For each profile, the mean household size and mean income band defined by the
ABCD Study are shown. The ABCD income range reflects the mean income band for a given profile. Based on 2024 U.S. Federal Poverty Guidelines issued
by the Department of Health and Human Services (https://aspe.hhs.gov/topics/poverty-economic-mobility/poverty-guidelines), income ranges for Profiles 3
and 5 fall below the federal poverty line for a household of 5 people. The federal poverty line for a household of 5 people is $36,580 in 48 contiguous states
and the District of Columbia. For the remaining profiles, the income ranges do not fall below the federal poverty line for their respective mean household
sizes. For each profile, the mean and range of the individual environmental factors also are shown.
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
18
Integrated Approach
The nine multi-system environmental profiles were dummy coded and Profile 1 was treated
as the reference profile as youth recorded average responses for the majority of the
environmental factors. First, we observed direct effects of the multi-system environmental
profiles on subcortical GM volume. Profile 2, i.e., family and neighborhood affluence during
late childhood, compared to Profile 1 predicted higher subcortical GM volume during early
adolescence (estimate [SE]=0.20 [0.059], P=0.001). Profile 3, i.e., family economic and
neighborhood adversity during late childhood, compared to Profile 1 predicted lower
subcortical GM volume during early adolescence (estimate [SE]=-0.43 [0.059], P<0.001).
Profile 5, i.e., adversity across, family, school, and neighborhood systems during late
childhood, compared to Profile 1 predicted lower subcortical GM volume during early
adolescence (estimate [SE]=-0.22 [0.094], P=0.018).
Next, we observed direct effects of the multi-system environmental profiles on externalizing
psychopathology. Profile 2, compared to Profile 1 predicted fewer externalizing problems
during early adolescence (estimate [SE]=-0.16 [0.060], P=0.009). Profile 5, compared to
Profile 1, predicted greater externalizing difficulties during early adolescence (estimate
[SE]=0.26 [0.095], P=0.006). Profile 9, i.e., family conflict and low school involvement
during late childhood, compared to Profile 1 predicted greater externalizing problems during
early adolescence (estimate [SE]=0.48 [0.14], P=0.001).
We also observed indirect effects of the multi-system environmental factors on externalizing
psychopathology via subcortical GM volume. The path Profile 2→Subcortical
2y→Externalizing 2y indicated that family and neighborhood affluence during late
childhood predicted higher subcortical GM volume during early adolescence, which in turn,
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
19
predicted fewer externalizing problems (estimate=-0.006, 95% bootstrap CI=[-0.020, -
0.00032]). The path Profile 3→Subcortical 2y→Externalizing 2y indicated that family
economic and neighborhood adversity during late childhood predicted lower subcortical GM
volume, which in turn, predicted greater externalizing difficulties during early childhood
(estimate=0.012, 95% bootstrap CI=[0.0012, 0.036]). See supplemental material for tests of
robustness by treating baseline subcortical GM volume and participant sex assigned at birth,
respectively (Figures S2-S3, available online).
Figure 3. Integrated approach linking multi-system environmental profiles, subcortical GM
volume, and externalizing/internalizing psychopathology. The integrated approach was
operationalized using a path analysis from structural equation modeling. The multi-system
environmental profiles (denoted by P2, …., P9) are derived from family, school, neighborhood, and
policy factors at baseline. The imaging measure corresponds to the subcortical gray matter (GM)
volume at 2-year follow-up. Externalizing and internalizing psychopathology are derived from the
Child Behavior Checklist (CBCL) at 2-year follow-up. Bold black arrows represent the significant
relationships (direct and indirect) with their respective coefficient estimates. Gray arrows indicate the
relationships between the multi-system environmental profiles and subcortical GM volume or
externalizing/internalizing psychopathology that are not significant. Circle arrows capture the
respective variances of subcortical GM volume, externalizing psychopathology, and internalizing
psychopathology. Circular line denotes the correlation between externalizing and internalizing
psychopathology. A bootstrapping procedure was performed to estimate the 95% confidence intervals
(CIs) of the indirect effects by selecting samples of youth randomly to perform the path analysis over
10,000 times. *P < 0.05. **P < 0.01. ***P < 0.001.
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
20
Discussion
The goal of the present study was to capture complex interactions among multiple
environmental systems and examine the impact of these systems on youth’s brain and
behavioral development as they transition from childhood to adolescence. To achieve this
goal, we first applied a novel Bayesian LPA framework to identify distinct subgroups of
children from family, school, neighborhood, and policy systems. We then assessed the ways
and degree to which the multi-system environmental profiles predicted subcortical GM
volume and externalizing and internalizing during early adolescence. The Bayesian LPA
revealed nine distinct profiles with excellent certainty and discrimination. The integrated
approach further captured direct and indirect influences of the multi-system environmental
profiles on subcortical GM volume and externalizing. Together, the transactions among
multi-system environments, brain structure, and psychopathology revealed an equifinality in
the ways subcortical volume and externalizing difficulties may be influenced during
adolescence.
Many foundational theories of development emphasize the importance of examining multiple
contexts at different levels of proximity to youth21,22,31,32. Some researchers take the approach
of documenting the additive accumulation of multiple environmental risks as it relates to
adolescent psychopathology33,34. Increasingly, though, researchers are identifying the co-
occurring interplay among different environmental experiences to identify relative
contributions of these environments on adolescent psychopathology35,36. A recent study using
the ABCD Study provided evidence of four distinct profiles of perceived threat across family,
school, and neighborhood systems35. These profiles reflected low threat across the three
systems, elevated threat in the neighborhood, elevated threat in the family, and elevated threat
across the three systems. Youth in the elevated threat across all systems profile had poorer
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
21
mental health outcomes, but youth in the family threat profile uniquely showed more
disruptive behavior symptoms and youth in the elevated neighborhood threat profile
displayed increased sleep problems. Another study found distinct groups of youth who
experienced low, medium, and high adversity, and maternal depression from family and
neighborhood systems36. Youth reported lowest externalizing and internalizing symptoms in
the low adversity followed by medium, maternal depression, and high adversity profiles.
These profile analyses are a great step in identifying the interplay among environmental
systems, and their impact on psychopathology. However, traditional LPA methods impose a
trade-off between number of participants in a sample and number of profiles that produces
meaningful discrimination—a large number of profiles leads to harder interpretation whereas
a smaller number of profiles leads to relative levels of low, medium, and high responses. Our
Bayesian LPA approach overcomes traditional limitations, revealing reliable profiles with
subtle variations in childhood environmental experiences24 that align with U.S. census data37
and highlight regional disparities in poverty and income (e.g., Profile 3 reflects the South,
with the highest childhood poverty and 1 in 5 Black women in poverty, while Profile 2
reflects the West, where White families dominate higher income levels). This approach
advances our understanding of how diverse environmental experiences interact and offers a
more nuanced framework for identifying risk and protective factors in adolescent
development.
We found three pathways from which late childhood multi-system environmental profiles
directly influenced subcortical volume. First, youth experiencing family and neighborhood
affluence predicted larger subcortical volume compared to youth in a below average family
income profile. Generally, environments that provide more opportunities and quality
resources can act as a buffer against stress for children and relate to better physical health
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
22
compared to those who belong to lower socioeconomic families and disadvantaged
neighborhoods38. Second, youth experiencing family economic and neighborhood adversity
had smaller subcortical volume compared to youth living in an environment with below
average family income. Third, adversity across, family, school, and neighborhood systems
during late childhood predicted smaller subcortical volume compared to lower family income
alone. The latter two effects are consistent with animal research showing that certain
subcortical regions are susceptible to effects of early adversities39. Early life adversity
negatively impacts neurogenesis (process of generating new neurons) in the hippocampus and
has been associated with increased dendritic arborization (process by which neurons create
connections with other neurons) in the amygdala39. In human research, children living in or
near poverty exhibit structural brain differences15,16, including smaller subcortical volumes.
Combined with previous research, the present study documents the relative effect that access
to fewer economic resources (at the family and neighborhood level) has on brain health. This
highlights the need to implement policies related to anti-poverty programs and neighborhood
resources (e.g., safety, amenities) to facilitate healthy developmental trajectories for youth.
The present analysis also revealed the unique ways different environmental systems directly
influence adolescent externalizing psychopathology. Youth who experienced relative family
and neighborhood affluence compared to youth living in families with lower income showed
fewer externalizing symptoms. However, youth who experienced adversity across family,
school, and neighborhood systems or who experienced family conflict and low school
involvement during late childhood had more externalizing problems in adolescence. It is
well-documented that youth who experience multi-system adversity show more externalizing
problems40,41. For instance, the family stress model posits that poverty, unsafe
neighborhoods, and economic instability stress parents, undermining their emotional
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
23
resources, and leading to greater family conflict, and eventually harsher parenting (and child
externalizing problems). This model has been supported by a host of empirical work42-44,
highlighting that broader environments play a role in shaping the proximal environment for
the developing youth. Notably, though, experiences of family conflict and low school
involvement reflected another robust risk pathway for externalizing, suggesting for some
youth adversity localized to family and school interactions confer risk for externalizing.
Parenting that is harsh and inconsistent has been linked robustly to the development of
externalizing behaviors45,46. This type of parenting is thought to model aggression for young
people, to undermine their ability to develop emotion regulation skills necessary for related
constructs such as empathy, and inconsistency in parenting is thought to lead to reward
contingencies that make aggression or breaking rules useful in some contexts. Further, youth
who experience conflict at home often bring this history to school where they begin these
types of cycles with peers and teachers, leading to trouble in school and often social rejection.
These findings underscore the need for researchers and clinicians to assess risk across
multiple systems. The conceptualization of a youth’s behavior should be based on the
combinations of risk factors that are influential for that person, providing a more personalized
and targeted approach to intervention.
Critically, the integrated path analysis highlighted key pathways whereby experiences within
environmental systems predicted externalizing problems via subcortical GM volume. Family
and neighborhood affluence during late childhood predicted higher subcortical GM volume
during early adolescence, which in turn, predicted fewer externalizing problems. This
pathway reveals how relative affluence buffers against externalizing psychopathology via
GM volume in brain regions responsible for reward, sensorimotor, cognitive, and emotion
processing. In contrast, family economic and neighborhood adversity predicted lower
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
24
subcortical GM volume, which in turn, predicted greater externalizing difficulties. This
finding is consistent with a previous study showing that subcortical GM trajectories mediate
the relationships between preschool socioeconomic status and high-risk behaviors15.
Therefore, there is growing evidence to show that socioeconomic adversities can negatively
impact structural brain development, which may increase the risk of mental health problems
through stress and lack of experiences that enrich development16,47. However, subcortical GM
volume did not mediate the relationship between the profile that described the most adverse
experiences across multiple systems (Profile 5) and externalizing problems. When a child
experiences multiple strong adversities, these adverse experiences provide a “push” to a
given outcome such that the importance of biological factors in these environments might be
diminished48. This is supported by the social push hypothesis48—when there are more and
stronger environmental pushes (e.g., harsh parenting in the context of neighborhood
disadvantage and crime) towards externalizing problems (such as aggression), biological risk
for aggression may matter less. Our findings support the consideration of an integrated
approach to parse the direct and indirect transactions between multiple environmental
systems, subcortical GM volume, and externalizing psychopathology.
Before concluding, it is important to note some limitations. First, we focused on subcortical
GM volume as subcortical structures are sensitive to socioeconomic resources and
psychopathology during childhood and adolescence1-5,13-16. However, other brain measures
also have been associated with socioeconomic resources49,50. Future work could identify
pathways that facilitate transactions between multiple environments, multi-modal brain
measures, and psychopathology. Second, while we used the ABCD Study as it samples the
sociodemographic variations longitudinally in the US population, our sample may not capture
the full spectrum of adversities that youth experience. As there was no relationship between
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
25
the multi-system environmental profiles and internalizing problems, future work could extend
the environmental factors that may confer risk for internalizing psychopathology. Finally, this
study follows a time lagged design to assess the ways and degree to which multi-system
environmental factors during late childhood predict subcortical volume and psychopathology
during early adolescence. We cannot infer the ways and extent to which subcortical volume
and psychopathology change longitudinally from childhood to adolescence. Future work
assessing how environmental systems influence baseline subcortical volume and
externalizing across different developmental stages could fully test the transactional nature of
environment-brain-psychopathology.
In conclusion, subcortical brain development and adolescent psychopathology are influenced
by multiple environmental systems related to family, school, neighborhood, and policy
factors. Our integrated approach highlights multiple equifinal pathways to adolescent
externalizing psychopathology, some directly via the environment and others via structural
brain development. Adverse environmental experiences should not just be viewed as
“challenges”, but instead as experiences that can shape the brain and behavior, ultimately
impacting mental health. Measuring environmental and neural factors could give us more
information about the status of particular risk/protective factors and help us refine our
understanding of who benefits from what interventions.
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
26
References
1. Peverill M, Dirks MA, Narvaja T, et al. Socioeconomic status and child psychopathology
in the United States: A meta-analysis of population-based studies. Clin Psychol Rev.
2021;83:101933. doi:10.1016/j.cpr.2020.101933
2. Jenson WR, Harward S, Bowen JM. Externalizing disorders in children and adolescents:
behavioral excess and behavioral deficits. In: Bray MA, Kehle TJ, eds. The Oxford
Handbook of School Psychology. Oxford Library of Psychology. Oxford Academic;
2011. doi:10.1093/oxfordhb/9780195369809.013.0141
3. Reijntjes A, Kamphuis JH, Prinzie P, et al. Peer victimization and internalizing problems
in children: a meta-analysis of longitudinal studies. Child Abuse Negl. 2010;34(4):244-
252. doi:10.1016/j.chiabu.2009.07.009
4. Aguilar-Yamuza B, Herruzo-Pino C, Lucena-Jurado V, et al. Internalizing Problems in
Childhood and Adolescence: The Role of the Family. Alpha Psychiatry. 2023;24(3):87-
92. doi:10.5152/alphapsychiatry.2023.221086
5. Miliauskas CR, Faus DP, da Cruz VL, et al. Community violence and internalizing
mental health symptoms in adolescents: A systematic review. BMC Psychiatry.
2022;22(1):253. doi:10.1186/s12888-022-03873-8
6. Estrada S, Gee DG, Bozic I, et al. Individual and environmental correlates of childhood
maltreatment and exposure to community violence: Utilizing a latent profile and a
multilevel meta-analytic approach. Psychol Med. 2023;53(1):189-205.
doi:10.1017/S0033291721001380
7. Quon EC, McGrath JJ. Subjective socioeconomic status and adolescent health: a meta-
analysis. Health Psychol. 2014;33(5):433-447. doi:10.1037/a0033716
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
27
8. Durham EL, Jeong HJ, Moore TM, et al. Association of gray matter volumes with
general and specific dimensions of psychopathology in children.
Neuropsychopharmacol. 2021;46:1333-1339. doi:10.1038/s41386-020-00952-w
9. Hyde LW, Bezek JL, Michael C. The future of neuroscience in developmental
psychopathology. Dev Psychopathol. 2024;1-16. doi:10.1017/S0954579424000233
10. LeDoux JE. Emotion circuits in the brain. Annu Rev Neurosci. 2000;23:155-184.
doi:10.1146/annurev.neuro.23.1.155
11. Mattoni M, Wilson S, Olino TM. Identifying profiles of brain structure and associations
with current and future psychopathology in youth. Dev Cogn Neurosci. 2021;51:101013.
doi:10.1016/j.dcn.2021.101013
12. Tooley UA, Bassett DS, Mackey AP. Environmental influences on the pace of brain
development. Nat Rev Neurosci. 2021;22(6):372-384. doi:10.1038/s41583-021-00457-5
13. Brito NH, Noble KG. Socioeconomic status and structural brain development. Front
Neurosci. 2014;8:276. doi:10.3389/fnins.2014.00276
14. Weissman DG, Hatzenbuehler ML, Cikara M, et al. State-level macro-economic factors
moderate the association of low income with brain structure and mental health in U.S.
children. Nat Commun. 2023;14(1):2085. doi:10.1038/s41467-023-37778-1
15. Barch DM, Donohue MR, Elsayed NM, et al. Early Childhood Socioeconomic Status
and Cognitive and Adaptive Outcomes at the Transition to Adulthood: The Mediating
Role of Gray Matter Development Across Five Scan Waves. Biol Psychiatry Cogn
Neurosci Neuroimaging. 2022;7(1):34-44. doi:10.1016/j.bpsc.2021.07.002
16. Maxwell MY, Taylor RL, Barch DM. Relationship Between Neighborhood Poverty and
Externalizing Symptoms in Children: Mediation and Moderation by Environmental
Factors and Brain Structure. Child Psychiatry Hum Dev. 2023;54(6):1710-1722.
doi:10.1007/s10578-022-01369-w
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
28
17. Cortes Hidalgo AP, Thijssen S, Delaney SW, et al. Harsh Parenting and Child Brain
Morphology: A Population-Based Study. Child Maltreat. 2022;27(2):163-173.
doi:10.1177/1077559520986856
18. Teicher MH, Anderson CM, Polcari A. Childhood maltreatment is associated with
reduced volume in the hippocampal subfields CA3, dentate gyrus, and subiculum. Proc
Natl Acad Sci U S A. 2012;109(9):E563-E572. doi:10.1073/pnas.1115396109
19. Baskin-Sommers A, Viding E, Barber M, et al. Advancing the science of biosocial
transactions related to aggression in children and young people: A brief review and steps
forward. Aggress Violent Behav. 2024;79:102001. doi:10.1016/j.avb.2024.102001
20. Hyde LW, Gard AM, Tomlinson RC, et al. An ecological approach to understanding the
developing brain: Examples linking poverty, parenting, neighborhoods, and the brain.
Am Psychol. 2020;75(9):1245-1259. doi:10.1037/amp0000741
21. Bronfenbrenner U. Toward an experimental ecology of human development. Am
Psychol. 1977;32(7):513–531. doi:10.1037/0003-066X.32.7.513
22. Eriksson M, Ghazinour M, Hammarström A. Different uses of Bronfenbrenner’s
ecological theory in public mental health research: What is their value for guiding public
mental health policy and practice? Soc Theory Health. 2018;16(4):414-433.
doi:10.1057/s41285-018-0065-6
23. Casey BJ, Cannonier T, Conley MI, et al. The Adolescent Brain Cognitive Development
(ABCD) study: Imaging acquisition across 21 sites. Dev Cogn Neurosci. 2018;32:43-54
doi:10.1016/j.dcn.2018.03.001
24. Paskewitz S, Brazil IA, Yildirim I, et al. Enhancing within-person estimation of
neurocognition and the prediction of externalizing behaviors in adolescents. Comput
Psychiatry. 2024;8(1):119-141. doi:10.5334/cpsy.112
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
29
25. Kim HH, McLaughlin KA, Chibnik LB, et al. Poverty, Cortical Structure, and
Psychopathologic Characteristics in Adolescence. JAMA Netw Open.
2022;5(11):e2244049. doi:10.1001/jamanetworkopen.2022.44049
26. Hagler DJ Jr, Hatton S, Cornejo MD, et al. Image processing and analysis methods for
the Adolescent Brain Cognitive Development Study. Neuroimage. 2019;202:116091.
doi:10.1016/j.neuroimage.2019.116091
27. Zucker RA, Gonzalez R, Feldstein Ewing SW, et al. Assessment of culture and
environment in the Adolescent Brain and Cognitive Development Study: Rationale,
description of measures, and early data. Dev Cogn Neurosci. 2018;32:107-120.
doi:10.1016/j.dcn.2018.03.004
28. Fan CC, Marshall A, Smolker H, et al. Adolescent Brain Cognitive Development
(ABCD) study Linked External Data (LED): Protocol and practices for geocoding and
assignment of environmental data. Dev Cogn Neurosci. 2021;52:101030.
doi:10.1016/j.dcn.2021.101030
29. Cardenas-Iniguez C, Schachner JN, Ip KI, et al. Building towards an adolescent neural
urbanome: Expanding environmental measures using linked external data (LED) in the
ABCD study. Dev Cogn Neurosci. 2024;65:101338. doi:10.1016/j.dcn.2023.101338
30. Achenbach TM, Rescorla LA. Manual for the ASEBA school-age forms & profiles: an
integrated system of multi-informant assessment.Burlington: University of Vermont,
Research Center for Children, Youth & Families; 2001
31. Lerner RM. Changing organism-context relations as the basic process of development: a
developmental contextual perspective. Dev Psychol. 1991;27(1):27-32.
doi:10.1037/0012-1649.27.1.27
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
30
32. Magnusson D, Stattin H. Person-context interaction theories. In: Damon W, Lerner RM,
eds. Handbook of Child Psychology: Theoretical Models of Human Development. 5th ed.
John Wiley & Sons, Inc.; 1998:685-759
33. Trentacosta CJ, Hyde LW, Shaw DS, et al. The relations among cumulative risk,
parenting, and behavior problems during early childhood. J Child Psychol Psychiatry.
2008;49(11):1211-1219. doi:10.1111/j.1469-7610.2008.01941.x
34. Evans GW, Li D, Whipple SS. Cumulative risk and child development. Psychol Bull.
2013;139(6):1342-1396. doi:10.1037/a0031808
35. Conley MI, Hernandez J, Salvati JM, et al. The role of perceived threats on mental
health, social, and neurocognitive youth outcomes: A multicontextual, person-centered
approach. Dev Psychopathol. 2023;35(2):689-710. doi:10.1017/S095457942100184X
36. Hardi FA, Beltz AM, McLoyd V, et al. Latent Profiles of Childhood Adversity,
Adolescent Mental Health, and Neural Network Connectivity. JAMA Netw Open.
2024;7(8):e2430711. doi:10.1001/jamanetworkopen.2024.30711
37. U.S. Census Bureau. American Community Survey: 5-year estimates. 2019. Available
from: https://data.census.gov
38. Roubinov DS, Hagan MJ, Boyce WT, et al. Family Socioeconomic Status, Cortisol, and
Physical Health in Early Childhood: The Role of Advantageous Neighborhood
Characteristics. Psychosom Med. 2018;80(5):492-501.
doi:10.1097/PSY.0000000000000585
39. Dufford AJ, Kim P, Evans GW. The impact of childhood poverty on brain health:
Emerging evidence from neuroimaging across the lifespan. Int Rev Neurobiol.
2020;150:77-105. doi:10.1016/bs.irn.2019.12.001
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
31
40. Supplee LH, Unikel EB, Shaw DS. Physical Environmental Adversity and the Protective
Role of Maternal Monitoring in Relation to Early Child Conduct Problems. J Appl Dev
Psychol. 2007;28(2):166-183. doi:10.1016/j.appdev.2006.12.001
41. Gautam N, Rahman MM, Khanam R. Adverse childhood experiences and externalizing,
internalizing, and prosocial behaviors in children and adolescents: A longitudinal
study. J Affect Disord. 2024;363:124-133. doi:10.1016/j.jad.2024.07.064
42. Conger RD, Donnellan MB. An interactionist perspective on the socioeconomic context
of human development. Annu Rev Psychol. 2007;58:175-199.
doi:10.1146/annurev.psych.58.110405.085551
43. Gard AM, McLoyd VC, Mitchell C, et al. Evaluation of a longitudinal family stress
model in a population-based cohort. Soc Dev. 2020;29(4):1155-1175.
doi:10.1111/sode.12446
44. McLoyd VC. The impact of economic hardship on Black families and children:
psychological distress, parenting, and socioemotional development. Child Dev.
1990;61(2):311-346. doi:10.1111/j.1467-8624.1990.tb02781.x
45. Masud H, Ahmad MS, Cho KW, et al. Parenting styles and aggression among young
adolescents: a systematic review of literature. Community Ment Health J. 2019;55:1015-
1030. doi:10.1007/s10597-019-00400-0
46. Reid JB, Patterson GR, Snyder J. Antisocial behavior in children and adolescents: a
developmental analysis and model for intervention. Washington, DC: American
Psychological Association; 2002. doi:10.1037/10468-000
47. Hyde LW, Gard AM, Tomlinson RC, et al. Parents, neighborhoods, and the developing
brain. Child Dev Perspect. 2022;16(3):148-156. doi:10.1111/cdep.12453
48. Raine A. Biosocial studies of antisocial and violent behavior in children and adults: A
review. J Abnorm Child Psychol.2002;30(4):311-326. doi:10.1023/A:1015754122318
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
32
49. Noble KG, Houston SM, Brito NH, et al. Family income, parental education and brain
structure in children and adolescents. Nat Neurosci. 2015;18(5):773-778.
doi:10.1038/nn.3983
50. Ip KI, Sisk LM, Horien C, et al. Associations among Household and Neighborhood
Socioeconomic Disadvantages, Resting-state Frontoamygdala Connectivity, and
Internalizing Symptoms in Youth. J Cogn Neurosci. 2022;34(10):1810-1841.
doi:10.1162/jocn_a_01826
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
33
Data Availability
The ABCD Study® is openly available following access permission granted to one or
multiple NIMH Data Archive (NDA) Collections (https://nda.nih.gov/nda/access-data-info).
The ABCD data repository grows and changes over time (https://nda.nih.gov/). The ABCD
data used in this report came from the tabulated data which can be navigated from the Data
Dictionary Explorer Release 5.1 (https://data-dict.abcdstudy.org/?). The detailed descriptions
of the assessments used in this report are available from the ABCD Study Protocols
(https://abcdstudy.org/scientists/protocols/).
Code Availability
The analysis code for the Bayesian LPA framework can be found at
https://github.com/SamPaskewitz/dpm.lpa. The analysis code for the integrated approach can
be found at https://github.com/JRam02/embeddedbrain.
Author Contributions Statement
J.R curated the demographic, environmental, behavioral, and imaging data from the ABCD
Study®. S.P developed the Bayesian LPA framework. J.R and S.P developed and tested the
integrated approach. J.R, S.P, and A.B.-S conceptualized this study, interpreted the analyses,
and wrote the original draft of the manuscript. J.R, S.P, I.A.B, and A.B.-S reviewed and
edited the manuscript. A.B.-S supervised this study.
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.17.628982doi: bioRxiv preprint
