Key PointsQuestion
Is neighborhood disadvantage associated with differences in youth neurocognition and brain structure after accounting for family socioeconomic status and does this association vary across US metropolitan areas?
Findings
In this cross-sectional study of 8598 children, neighborhood disadvantage was associated with worse neurocognitive performance and with lower total cortical surface area and subcortical volume. These associations were similar across the United States and were attributed to local differences in neighborhood disadvantage within metropolitan areas.
Meaning
Findings from this study suggest that local variations in neighborhood disadvantage are an environmental risk factor for youth neurocognitive performance and brain structure across the US, and thus improving the neighborhood context may be a promising approach to achieving better short- and long-term health and development for children and adolescents.
Importance
Neighborhood disadvantage is an important social determinant of health in childhood and adolescence. Less is known about the association of neighborhood disadvantage with youth neurocognition and brain structure, and particularly whether associations are similar across metropolitan areas and are attributed to local differences in disadvantage.
Objective
To test whether neighborhood disadvantage is associated with youth neurocognitive performance and with global and regional measures of brain structure after adjusting for family socioeconomic status and perceptions of neighborhood characteristics, and to assess whether these associations (1) are pervasive or limited, (2) vary across metropolitan areas, and (3) are attributed to local variation in disadvantage within metropolitan areas.
Design, Setting, and Participants
This cross-sectional study analyzed baseline data from the Adolescent Brain and Cognitive Development (ABCD) Study, a cohort study conducted at 21 sites across the US. Participants were children aged 9.00 to 10.99 years at enrollment. They and their parent or caregiver completed a baseline visit between October 1, 2016, and October 31, 2018.
Exposures
Neighborhood disadvantage factor based on US census tract characteristics.
Main Outcomes and Measures
Neurocognition was measured with the NIH Toolbox Cognition Battery, and T1-weighted magnetic resonance imaging was used to assess whole-brain and regional measures of structure. Linear mixed-effects models examined the association between neighborhood disadvantage and outcomes after adjusting for sociodemographic factors.
Results
Of the 11 875 children in the ABCD Study cohort, 8598 children (72.4%) were included in this analysis. The study sample had a mean (SD) age of 118.8 (7.4) months and included 4526 boys (52.6%). Every 1-unit increase in the neighborhood disadvantage factor was associated with lower performance on 6 of 7 subtests, such as Flanker Inhibitory Control and Attention (unstandardized Β = −0.5; 95% CI, −0.7 to −0.2; false discovery rate (FDR)–corrected P = .001) and List Sorting Working Memory (unstandardized Β = −0.7; 95% CI, −1.0 to −0.3; FDR-corrected P < .001), as well as on all composite measures of neurocognition, such as the Total Cognition Composite (unstandardized Β = −0.7; 95% CI, −0.9 to −0.5; FDR-corrected P < .001). Each 1-unit increase in neighborhood disadvantage was associated with lower whole-brain cortical surface area (unstandardized Β = −692.6 mm2; 95% CI, −1154.9 to −230.4 mm2; FDR-corrected P = .007) and subcortical volume (unstandardized Β = −113.9 mm3; 95% CI, −198.5 to −29.4 mm3; FDR-corrected P = .03) as well as with regional surface area differences, primarily in the frontal, parietal, and temporal lobes. Associations largely remained after adjusting for perceptions of neighborhood safety and were both consistent across metropolitan areas and primarily explained by local variation in each area.
Conclusions and Relevance
This study found that, in the US, local variation in neighborhood disadvantage was associated with lower neurocognitive performance and smaller cortical surface area and subcortical volume in young people. The findings demonstrate that neighborhood disadvantage is an environmental risk factor for neurodevelopmental and population health and enhancing the neighborhood context is a promising approach to improving the health and development of children and adolescents.
Neighborhood disadvantage is an important social determinant of physical and mental health in childhood and adolescence, independent of family socioeconomic status (SES).1-9 A growing body of work has demonstrated the association of family SES with brain structure and neurocognition,10-14 but less is known about the role of neighborhood disadvantage. Neighborhood disadvantage may operate as a broader social determinant of these outcomes, potentially contributing to the widening health inequality across the life span, and act as a potential target for prevention.15-17
Although neighborhood disadvantage has been consistently linked to broad measures of cognition,4,18-22 limited evidence exists regarding its associations with specific neurocognitive domains.23-28 Although neighborhood disadvantage has been associated with worse language performance,23 there has been mixed evidence of its associations with executive functioning and working memory.24-26 Studies with large samples have found pervasive differences across neurocognitive domains but were unable to control for both family income and parental educational level to disentangle the role of family and neighborhood factors.27,28
Neighborhood disadvantage has been associated with total gray matter volume without accounting for family SES,27 although another study found it was associated with cortical thickness and increased amygdala volume and thickening in the temporal lobe while controlling for family SES.29 Recent studies have found associations with prefrontal cortex structure and hippocampal volume but did not consider the broader pattern of whole-brain and regional differences.22,28 Consequently, studies are needed to assess neighborhood-related differences in whole-brain and regional patterns of neurodevelopment while accounting for family SES.
In addition, it has not been ascertained whether objectively measured neighborhood disadvantage has a distinct role, apart from the perceptions of neighborhood social characteristics, that may affect child development by different mechanisms.30,31 The evidence is mixed on whether perceptions of physical disorder and social cohesion are associated with cognition,32-36 whereas perceptions of safety37 may be particularly important given the association between community violence and neurocognition38-42 and limbic system volume.43,44
No studies have assessed whether the associations between neighborhood disadvantage, neurocognition, and neurodevelopment vary across metropolitan areas in the United States.30,45,46 The meaning and correlates of neighborhood disadvantage across cities are heterogeneous, and thus variation exists in potential mechanisms of neighborhood effects.46-48 Studies have also found evidence that neighborhood disparities vary by geographic location for some cognitive and developmental outcomes49-51 and that the implications of experimental changes in children’s neighborhoods vary across cities.52,53 These findings raise the questions of whether the association between neighborhood disadvantage, neurocognition, and brain structure varies across the US and whether the associations can be attributed to the overall differences in characteristics between metropolitan areas or the local variation within each metropolitan area. These alternatives have critical public health relevance in terms of pervasiveness of such associations and whether the potential mechanisms are the types that differentiate neighborhoods within each local area.
These questions are addressed in the Adolescent Brain Cognitive Development (ABCD) Study, a large cohort of children aged 9 to 10 years from 21 study sites throughout the US. To our knowledge, the present cross-sectional study of the ABCD Study cohort is the first large-scale test of the hypothesis that neighborhood disadvantage is associated with youth neurocognitive performance and with global and regional measures of brain structure after adjusting for family SES and perceptions of neighborhood characteristics. In addition, to date, this study is the first to assess whether associations with neighborhood disadvantage (1) are pervasive or limited, (2) vary across metropolitan areas, and (3) are attributed to local variation in disadvantage within metropolitan areas.
This study received approval from the institutional review board of the University of Southern California. The ABCD Study obtained centralized institutional review board approval from the University of California, San Diego, and each of the 21 study sites obtained local institutional review board approval. Ethical regulations were followed during data collection and analysis. Parents or caregivers provided written informed consent, and children gave written assent.
Participants and Procedures
Data were obtained from the baseline assessment of the ABCD Study (2019 National Institute of Mental Health Data Archive 2.0.1 release), a large cohort study conducted across 21 US metropolitan areas (Children’s Hospital Los Angeles, Los Angeles, California; Florida International University, Miami, Florida; Laureate Institute for Brain Research, Tulsa, Oklahoma; Medical University of South Carolina, Charleston, South Carolina; Oregon Health and Science University, Portland, Oregon; SRI International, Menlo Park, California; University of California San Diego, San Diego, California; UCLA [University of California, Los Angeles, California]; University of Colorado Boulder, Boulder, Colorado; University of Florida, Gainesville, Florida; University of Maryland at Baltimore, Baltimore, Maryland; University of Michigan, Ann Arbor, Michigan; University of Minnesota, Minneapolis, Minnesota; University of Pittsburgh, Pittsburgh, Pennsylvania; University of Rochester, Rochester, New York; University of Utah, Salt Lake City, Utah; University of Vermont, Burlington, Vermont; University of Wisconsin-Milwaukee, Milwaukee, Wisconsin; Virginia Commonwealth University, Richmond, Virginia; Washington University in St. Louis, St. Louis, Missouri; and Yale University, New Haven, Connecticut).54 Participants were recruited through a stratified probability sampling for each site at the school level (eMethods in the Supplement).55,56 Children were aged 9.00 to 10.99 years at enrollment, and they and their parent or caregiver completed a baseline visit between October 1, 2016, and October 31, 2018. The visit consisted of clinical interviews, surveys, neurocognitive tests, and neuroimaging.55-63
We excluded participants with a nonvalid address and randomly selected 1 family member for inclusion to reduce nonindependence (eMethods in the Supplement). For neurocognition analyses, we excluded participants with missing cognitive, sociodemographic, and/or neighborhood data. For neuroimaging analyses, we further excluded participants whose magnetic resonance imaging scans did not pass quality control or displayed incidental findings. The participant selection flowchart is shown in eFigure 1 in the Supplement, and participant characteristics are listed in Table 1; details regarding excluded participants are in eMethods and eTable 1 in the Supplement.
Neighborhood Disadvantage
The child’s primary residential address at baseline was geocoded by the Data Analysis, Informatics and Resource Center of the ABCD Study, and variables from the American Community Survey (5-year estimates from 2011 to 2015) were linked to each individual according to their US census tract. We selected 5 nonredundant constructs frequently used in metrics of disadvantage23,64-67 that are not dependent on real estate markets: percentage of residents with at least a high school diploma, median family income, unemployment rate, percentage of families living below the federal poverty level, and percentage of single-parent households (Table 1; eTable 2 in the Supplement). We created a single neighborhood disadvantage factor score using a maximum likelihood exploratory factor analysis that explained 67% of the variance (eTable 2 in the Supplement).
Neurocognitive Assessment and Neuroimaging
Neurocognitive performance was measured using the NIH Toolbox Cognition Battery, specifically the Dimensional Change Card Sort, Flanker Inhibitory Control and Attention, List Sorting Working Memory, Oral Reading Recognition, Pattern Comparison Processing Speed, Picture Sequence Memory, and Picture Vocabulary tests.58,68-70 The performance on these 7 tests is summarized in the composite scores, including a total cognitive score as well as crystallized and fluid cognition scores.68,71 Uncorrected standard scores were used as primary dependent variables (eTable 1 in the Supplement).
As described previously, magnetic resonance imaging methods and assessments were optimized and harmonized across ABCD Study sites for 3-T scanners (Discovery MR750, GE Healthcare; Achieva dStream and Ingenia, Philips Healthcare; Prisma and Prisma Fit, Siemens Medical Solutions).60,63 Cortical surface reconstruction and subcortical segmentation were processed through FreeSurfer, version 5.3.0 (FreeSurfer), using the T1-weighted anatomical scans, including total gray and white matter as well as subcortical volumes (cubic millimeter), cortical thickness (millimeter), and cortical surface area (square millimeter) estimates for cortical regions using the Desikan-Killiany Atlas.63,72,73 The Data Analysis, Informatics and Resource Center of the ABCD Study performed quality control procedures to estimate the severity of motion, intensity inhomogeneity, white matter underestimation, pial overestimation, and magnetic susceptibility of the artifact.63
Family Socioeconomic Status and Neighborhood Perceptions
Parental educational level was the highest educational achievement by either parent or caregiver. Total family income covered all sources of income for family members, including wages, benefits, child support payments, and others. It was assessed in ordinal ranges, and we used the midpoint of the range divided by $10 000 to create and scale a continuous variable. Subjective perceptions of neighborhood safety were rated for 3 items by parents and for 1 item by children (eMethods in the Supplement).62,74,75
Linear mixed-effects models were used to estimate the association of neighborhood disadvantage with neurocognitive performance and with whole-brain measures. Site-level random intercepts were used to accommodate correlation from clustering of individuals within study sites. Models included age, sex assigned at birth, race/ethnicity, parental educational level, and family income as covariates. Magnetic resonance imaging analyses also included the device manufacturer, handedness,76 and intracranial volume for volumetric analyses. Incorporating neighborhood disadvantage improved the model fit in all cases in which outcomes were significantly associated with neighborhood disadvantage (eTable 5 in the Supplement). Additional models were then fitted to (1) control for subjective perceptions of neighborhood safety, (2) test whether heterogeneity existed in the parameters for neighborhood disadvantage across sites by adding a random slope, and (3) test whether the associations were attributed to the site differences in neighborhood disadvantage level (between-site variation, a site mean score) vs relative local variations in neighborhood disadvantage (within-site variation, an individual’s relative deviation from the site mean score).77 As a follow-up to the significant associations between whole-brain measures and neighborhood disadvantage, we conducted a region of interest (ROI) analysis.
Tests of significance (2-tailed) were corrected for multiple comparisons using a false discovery rate (FDR) correction, with P < .05 as the corrected threshold for significance.78 All analyses were performed using the nlme, lmertest, sjstats, dplyr, psych, and ggplot2 packages as well as the factanal function for factor analysis in R, version 4.0 (R Foundation for Statistical Computing).
Both unstandardized and standardized parameters were reported for all models to aid in the interpretation of findings. Unstandardized Β corresponded to the increase or decrease in the outcome, using its original scale, for a 1-unit change in the neighborhood disadvantage factor. Table 1 illustrates the mean and SD for the overall neighborhood disadvantage factor score as well as for within-site and between-site variations in neighborhood disadvantage. Standardized β corresponded to the increase or decrease of the outcome, in SDs of its distribution, for every 1-SD change in the neighborhood disadvantage scores. The eMethods in the Supplement contains additional details regarding the statistical analyses.
Of the 11 875 children in the ABCD Study cohort, we included 8598 children (72.4%) for neurocognition analyses and 7650 children (64.4%) for neuroimaging analyses. The study sample comprised 4526 boys (52.6%) and 4072 girls (47.4%) with a mean (SD) age of 118.8 (7.4) months (Table 1).
Significant differences in neighborhood disadvantage were observed across study sites (F1,21 = 78.6; P < .001) (eFigure 2 in the Supplement). Higher neighborhood disadvantage was associated with lower family income, lower parental and child perceptions of neighborhood safety, race/ethnicity, and parental educational level (eTables 3 and 4 in the Supplement).
Neighborhood Disadvantage and Neurocognitive Performance
Higher neighborhood disadvantage had an inverse association with 6 of the 7 neurocognitive subtests. Specifically, a 1-unit increase in the neighborhood disadvantage factor score was associated with lower scores on the following measures: Flanker Inhibitory Control and Attention (unstandardized Β = −0.5; 95% CI, −0.7 to −0.2; FDR-corrected P = .001), List Sorting Working Memory (unstandardized Β = −0.7; 95% CI, −1.0 to −0.3; FDR-corrected P < .001), Dimensional Change Card Sort (unstandardized Β = −0.4; 95% CI, −0.7 to −0.2; FDR-corrected P = .003), Oral Reading Recognition (unstandardized Β = −0.4; 95% CI, −0.6 to −0.2; FDR-corrected P < .001), Pattern Comparison Processing Speed (unstandardized Β = −0.6; 95% CI, −1.0 to −0.2; FDR-corrected P = .009), and Picture Vocabulary (unstandardized Β = −0.7; 95% CI, −0.9 to −0.5; FDR-corrected P < .001) as well as all composite measures (Figure 1 and Table 2). Standardized β parameters for neighborhood disadvantage were approximately 60% to 90% of those for family income and were smaller than that of a parental postgraduate degree (eTable 6 in the Supplement). In addition, as shown in Figure 1, considerable individual variability was found, and many from disadvantaged neighborhoods outperformed their peers from more affluent neighborhoods. Of the 9 associations between neighborhood disadvantage and cognition, all remained significant when perceptions of safety were included, except for processing speed (eTable 7 in the Supplement). Follow-up analyses, in which a random slope for neighborhood disadvantage was added, revealed little evidence that associations between neighborhood disadvantage and cognition were different across sites (eTable 8 in the Supplement).
Associations for within-site variation in neighborhood disadvantage were largely similar to those for the overall neighborhood disadvantage factor (Table 2). Specifically, a 1-unit increase in the neighborhood disadvantage factor score was associated with lower scores on the following measures: Flanker Inhibitory Control and Attention (unstandardized Β = −0.4; 95% CI, −0.7 to −0.2; FDR-corrected P = .001), List Sorting Working Memory (unstandardized Β = −0.6; 95% CI, −0.9 to −0.3; FDR-corrected P < .001), Dimensional Change Card Sort (unstandardized Β = −0.4; 95% CI, −0.6 to −0.1; FDR-corrected P = .006), Oral Reading Recognition (unstandardized Β = −0.4; 95% CI, −0.6 to −0.2; FDR-corrected P < .001), Pattern Comparison Processing Speed (unstandardized Β = −0.5; 95% CI, −0.9 to −0.1; FDR-corrected P = .01), and Picture Vocabulary (unstandardized Β = −0.7; 95% CI, −0.9 to −0.5; FDR-corrected P < .001) as well as all composite measures (Figure 1 and Table 2). Between-site differences in neighborhood disadvantage were solely associated with Flanker Inhibitory Control and Attention (unstandardized Β = −1.3; 95% CI, −2.2 to −0.4; FDR-corrected P = .04) and Fluid Cognition Composite (unstandardized Β = −1.7; 95% CI, −2.9 to −0.5; FDR-corrected P = .04) (Table 2).
Neighborhood Disadvantage and Brain Structure
Greater neighborhood disadvantage was associated with whole-brain structure. Specifically, a 1-unit increase in the neighborhood disadvantage factor score was associated with smaller total cortical surface area (unstandardized Β = −692.6 mm2; 95% CI, −1154.9 to −230.4 mm2; FDR-corrected P = .007), cortical gray matter volume (unstandardized Β = −892.1 mm3; 95% CI, −1679.2 to −105.0 mm3; FDR-corrected P = .04), and subcortical gray matter volume (unstandardized Β = −113.9 mm3; 95% CI, −198.5 to −29.4 mm3; FDR-corrected P = .03) but not with cortical thickness or white matter volume (Figure 2A and Table 3). When comparing standardized parameters, the association of neighborhood disadvantage with surface area was approximately two-thirds the size of that for family income, and smaller than parental educational level, although it was larger than both for subcortical volume (eTable 6 in the Supplement). In addition, as shown in Figure 2, considerable individual variability was found, and many from disadvantaged neighborhoods had larger cortical surface area, cortical gray matter volume, and subcortical volume than their peers from more affluent neighborhoods. Perceptions of neighborhood safety were not associated with any whole-brain measure (eTable 9 in the Supplement). The association between neighborhood disadvantage and smaller cortical surface area remained after controlling for subjective perceptions, although the estimates were attenuated by 18.5% for cortical gray matter volume and by 10.3% for subcortical volume and were no longer significant (eTable 9 in the Supplement). The variance for the random slope for neighborhood disadvantage was not significant for any whole-brain measure, and thus no evidence was found that the associations between neighborhood disadvantage and whole-brain structure were different across sites (eTable 8 in the Supplement).
The magnitude, direction, and associations for within-site variation in neighborhood disadvantage were largely similar to those for overall neighborhood disadvantage. Specifically, a 1-unit increase in the neighborhood disadvantage factor score was associated with smaller total cortical surface area (unstandardized Β = −707.6 mm2; 95% CI, −1173.3 to −241.9 mm2; FDR-corrected P = .006), cortical gray matter volume (unstandardized Β = −903.1 mm3; 95% CI, −1692.8 to −113.5 mm3; FDR-corrected P = .04), and subcortical gray matter volume (unstandardized Β = −121.0 mm3; 95% CI, −206.2 to −35.7 mm3; FDR-corrected P = .02). No associations with between-site variation in neighborhood disadvantage were found (Table 3).
Region of Interest Analyses
Results of cortical surface area ROI analyses are shown in Figure 2B and eTable 10 in the Supplement. Greater neighborhood disadvantage was associated with smaller surface areas in 9 ROIs across all lobes of the brain: rostral middle frontal (unstandardized Β = −33.4 mm2; 95% CI, −56.6 to −10.2 mm2; FDR-corrected P = .03), pars orbitalis (unstandardized Β = −4.7 mm2; 95% CI, −7.2 to −2.2 mm2; FDR-corrected P = .005), precentral (unstandardized Β = −30.9 mm2; 95% CI, −47.2 to −14.6 mm2; FDR-corrected P = .005), superior parietal (unstandardized Β = −35.0 mm2; 95% CI, −54.6 to −15.3 mm2; FDR-corrected P = .006), precuneus (unstandardized Β = −21.3 mm2; 95% CI, −36.1 to −6.5 mm2; FDR-corrected P = .04), inferior temporal (unstandardized Β = −16.6 mm2; 95% CI, −29.7 to −3.5 mm2; FDR-corrected P = .04), entorhinal (unstandardized Β = −3.7 mm2; 95% CI, −5.9 to −1.5 mm2; FDR-corrected P = .01), fusiform (unstandardized Β = −15.6 mm2; 95% CI, −27.0 to −4.1 mm2; FDR-corrected P = .04), and cuneus (unstandardized Β = −7.6 mm2; 95% CI, −13.5 to −1.7 mm2; FDR-corrected P = .03). No associations with between-site variation were found, and the patterns observed for within-site variation in neighborhood disadvantage were similar to the overall pattern (Figure 2B and eTable 10 in the Supplement). When controlling for total cortical surface area, ROIs were not associated with neighborhood disadvantage (eTable 11 in the Supplement).
No associations between neighborhood disadvantage and subcortical volumes were observed for any ROI, nor were associations found for between-site or within-site variations in neighborhood disadvantage (eTable 13 in the Supplement). Analyses of cortical volume followed a similar but less robust pattern as surface area (eResults, eFigure 3, and eTable 12 in the Supplement).
Neighborhood disadvantage was associated with worse neurocognitive performance and with smaller cortical surface area as well as cortical volumes and subcortical volumes. These associations remained after adjusting for family SES or largely remained after adjusting for perceptions of neighborhood safety. Thus, these social disparities merit further study to assess their prospective role in increasing developmental disparities in health and mental health15-17,79 and in the differences in neurocognition and brain structure in adulthood.80-85 These associations were largely consistent across 21 US metropolitan areas, suggesting a widespread and potentially generalizable pattern. Moreover, associations were primarily attributed to the local variation in neighborhood disadvantage within metropolitan areas.
The association between neighborhood disadvantage and performance was found for cognitive control/attention, working memory, flexible thinking, reading ability, and language as well as all composite measures. This finding suggests that disparities in overall cognitive performance4,18-22 overlay pervasive differences across nearly all specific neurocognitive domains independent of family SES10,11,86,87 and perceptions of safety. Moreover, the broad pattern suggests that underlying mechanisms may be general and prevention approaches may be most successful if they are comprehensive rather than narrowly targeted to the development of particular cognitive skills.
Neighborhood disadvantage was also associated with global differences in brain structure after accounting for family income and parental educational level,11-13 and differences in cortical surface area remained after adjusting for perceptions of safety. This finding is consistent with similar associations after adjustment for family SES found in functional neuroimaging studies.88-90 Although neighborhood disadvantage was also associated with cortical surface area in frontal, parietal, and temporal lobe regions, these regions were accounted for by global differences in surface area. This finding suggests that interpretations of uncorrected regional differences,22,27,29 although still potentially meaningful for developmental outcomes,91 are incomplete without contextualization within the broader global pattern across the brain.
Cortical surface area exhibits nonlinear growth and decreases across adolescence,92 with lobar variation in the trajectory and peak of growth.92-95 Consequently, the inverse associations in this study were consistent with the stress acceleration hypothesis, which proposes that early adversity will be associated with accelerated maturation in neural regions, particularly those regions related to stress and emotion.96 The inverse association with total subcortical volume, with nonsignificant differences in specific regions, was surprising because both childhood disadvantage11,28,97,98 and neighborhood disadvantage in adulthood84 have been associated with hippocampal volume. However, regionally specific associations may depend on demographic controls or may emerge with development given that growth in the hippocampus is nonlinear99,100 and disadvantage29,101 has been associated with change in subcortical structure.
Although the magnitudes of association were statistically small, they are potentially meaningful. First, small effect sizes may have large consequences because they accumulate over time at a population level.102 This theory is particularly salient for neighborhoods given that both early and cumulative exposures may be particularly influential.45 Second, the magnitudes are comparable to but smaller than the effect sizes for family SES in these models, given that the strength of associations was typically about 60% to 90% of that for family income for most measures, which has more well-established associations with brain and neurocognitive development.10-14,103 Nevertheless, these estimates may be conservative because they included all SES measures in the same model. However, these associations were not risk factors at the individual level. Many children and adolescents from disadvantaged neighborhoods outperformed their peers from more affluent neighborhoods, and they had larger cortical surface area and subcortical volume as well (as noted by the individual variability shown in Figures 1 and 2).
These findings have implications for the potential mechanisms of possible neighborhood effects. First, neighborhood disadvantage was more consistently associated with diverse outcomes than were perceptions of safety.30,31,45 Second, the similarity of neighborhood effects across the country and the prominent role of within-region variation suggest that the most plausible mechanisms are social factors that are often associated with neighborhood disadvantage, such as community resources and services, schools, nutritional environments or walkability, environmental pollutants, community violence, and green space.2-4,38,43,103-105 An additional possibility is that neighborhood disadvantage may operate as a fundamental cause of health outcomes, with social mechanisms that may vary across local contexts.106 Moreover, living in a disadvantaged neighborhood functions as a chronic stressor.107 Neighborhood disadvantage has been associated with hypothalamic-pituitary-adrenal axis function and allostatic load,108-111 with the hypothalamic-pituitary-adrenal axis implicated as a mediator in adults,85 suggesting that this theory is plausible. Future longitudinal work should investigate whether such factors may serve as mediators of the associations with neighborhood disadvantage over time.
Despite these implications, the cross-sectional design of this study limits causal inference. However, the results are consistent with those reported in experimental and quasi-experimental studies as well as polygenic risk analyses, which were supportive of the causal association between neighborhood disadvantage and the physical and mental health of youth.9,21,53,112-114 Genetically informed studies may help to clarify the environmental mechanisms that can transmit such inequities.115 There is no evidence that such neighborhood-related differences are fixed or immutable. Evidence points to the potential plasticity and malleability related to social and contextual interventions that improve youth environments.103 Consequently, prevention and intervention strategies must include a focus on improving neighborhood environments at the local metropolitan level, rather than only considering individual-level factors, to narrow disparities among youth.
The convergence of findings across both neurocognition and brain structure, along with interrelated trajectories of change in neurocognitive and structural brain development,92,95,100,116-119 prompts the speculation that differences in brain structure may partially mediate neurocognitive disparities. This result has been observed for family SES.11,120 However, given that cross-sectional mediation analyses may be biased,121 these questions are best addressed in future longitudinal studies.
This study has several limitations. First, ABCD Study data restrictions do not allow for consideration of clustering or spatial relationships among neighborhoods, and thus unmeasured neighborhood-level factors cannot be fully addressed. Second, caution is warranted in comparing effect sizes for between-site and within-site variations in neighborhood disadvantage, in part because within-site variation in disadvantage was greater and likely captured different constructs compared with the between-site variation. Nevertheless, in the few cases in which the mean differences in neighborhood disadvantage between sites were associated with neurocognition, the associations were similar or even larger in size. Third, because the study sample had less neighborhood disadvantage than the US overall and more disadvantaged participants were more likely to be excluded because of missing data, associations with neighborhood disadvantage may be underestimated.
To our knowledge, this study was the first large, multisite study to find that neighborhood disadvantage was associated with a pervasive pattern of worse neurocognitive performance and with both whole-brain and regional differences in brain structure, after controlling for family income and parental educational level as well as subjective neighborhood perceptions. These neighborhood associations were largely consistent across the US, were attributed to local variation in neighborhood disadvantage, and were consistent with previous reports of an association of neighborhood disadvantage with physical and mental health in childhood and adolescence. The findings suggest that neighborhood disadvantage is a central environmental risk factor for neurodevelopmental and population health in the US and that enhancing the neighborhood context may improve the short- and long-term health and development of children and adolescents.
Accepted for Publication: February 1, 2021.
Published Online: May 3, 2021. doi:10.1001/jamapediatrics.2021.0426
Corresponding Authors: Daniel A. Hackman, PhD, USC Suzanne Dworak-Peck School of Social Work, University of Southern California, 669 W 34th St, Los Angeles, CA 90089 (dhackman@usc.edu); Megan M. Herting, PhD, Department of Preventive Medicine, Keck School of Medicine of University of Southern California, 2001 N Soto St, Room 225N, Los Angeles, CA 90033 (herting@usc.edu).
Author Contributions: Dr Hackman and Ms Cserbik had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Hackman, McConnell, Herting.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Hackman, Cserbik, Berhane.
Critical revision of the manuscript for important intellectual content: Cserbik, Chen, Berhane, Minaravesh, McConnell, Herting.
Statistical analysis: Cserbik, Chen, Berhane.
Obtained funding: Chen, Herting.
Supervision: Hackman, Berhane, Herting.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was partially supported by National Institute of Environmental Health Sciences grants P30ES007048-23S1 (supported Ms Cserbik), P01ES022845 (PI: Dr McConnell; also supported Drs Herting, McConnell, Chen, and Berhane), and R01ES031074 (PI: Dr Herting; also funded Drs Hackman, McConnell, Chen, and Berhane) from the National Institutes of Health (NIH). Dr Herting and Ms Cserbik were supported by the Rose Hills Foundation. Dr Hackman was supported by grant R21HD099596 from the NIH. The Adolescent Brain Cognitive Development (ABCD) Study was supported by the NIH and other 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, and U24DA041147. A full list of supporters is available at https://abcdstudy.org/federal-partners.html. A listing of participating sites and a complete listing of the study investigators can be found at https://abcdstudy.org/consortium_members/. The ABCD consortium investigators designed and implemented the study and/or provided data but did not necessarily participate in analysis or writing of this manuscript.
Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Disclaimer: The views expressed herein are those of the authors and do not reflect the official policy or position of the NIH or ABCD consortium investigators.
Additional Contributions: Woo Jung Lee, MA, USC Suzanne Dworak-Peck School of Social Work, University of Southern California, provided assistance with tables. Abigail Palmer Molina, MA, LCSW, IFECMHS, PMH-C, USC Suzanne Dworak-Peck School of Social Work, University of Southern California, provided assistance with literature review. These individuals did not receive specific compensation for their contributions beyond their support as research assistants.
Additional Information: Data used in the preparation of this article were obtained from the ABCD Study (https://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 data repository grows and changes over time. The ABCD data used in this report came from NIMH Data Archive Digital Object Identifier (https://dx.doi.org/10.15154/1504041). The NDA study for this project can be found at https://dx.doi.org/10.15154/1519209.
14.Judd
N, Sauce
B, Wiedenhoeft
J,
et al. Cognitive and brain development is independently influenced by socioeconomic status and polygenic scores for educational attainment.
Proc Natl Acad Sci U S A. 2020;117(22):12411-12418. doi:
10.1073/pnas.2001228117
PubMedGoogle ScholarCrossref 15.Shonkoff
JP, Boyce
WT, McEwen
BS. Neuroscience, molecular biology, and the childhood roots of health disparities: building a new framework for health promotion and disease prevention.
JAMA. 2009;301(21):2252-2259. doi:
10.1001/jama.2009.754
PubMedGoogle ScholarCrossref 20.Moore
TM, Martin
IK, Gur
OM,
et al. Characterizing social environment’s association with neurocognition using census and crime data linked to the Philadelphia Neurodevelopmental Cohort.
Psychol Med. 2016;46(3):599-610. doi:
10.1017/S0033291715002111
PubMedGoogle ScholarCrossref 21.Engelhardt
LE, Church
JA, Paige Harden
K, Tucker-Drob
EM. Accounting for the shared environment in cognitive abilities and academic achievement with measured socioecological contexts.
Dev Sci. 2019;22(1):e12699. doi:
10.1111/desc.12699
PubMedGoogle Scholar 24.Umbach
R, Raine
A, Gur
RC, Portnoy
J. Neighborhood disadvantage and neuropsychological functioning as part mediators of the race–antisocial relationship: a serial mediation model.
J Quant Criminol. 2018;34:481-512. doi:
10.1007/s10940-017-9343-z
Google ScholarCrossref 32.Choi
JK, Kelley
MS, Wang
D. Neighborhood characteristics, maternal parenting, and health and development of children from socioeconomically disadvantaged families.
Am J Community Psychol. 2018;62(3-4):476-491. doi:
10.1002/ajcp.12276
PubMedGoogle ScholarCrossref 33.Fishbein
DH, Michael
L, Guthrie
C, Carr
C, Raymer
J. Associations between environmental conditions and executive cognitive functioning and behavior during late childhood: a pilot study.
Front Psychol. 2019;10:1263. doi:
10.3389/fpsyg.2019.01263
PubMedGoogle ScholarCrossref 34.St. John
AM, Tarullo
AR. Neighbourhood chaos moderates the association of socioeconomic status and child executive functioning.
Infant Child Dev. 2020;29(1): e2153. doi:
10.1002/icd.2153
Google Scholar 36.Zaheed
AB, Sharifian
N, Kraal
AZ, Sol
K, Hence
A, Zahodne
LB. Unique effects of perceived neighborhood physical disorder and social cohesion on episodic memory and semantic fluency.
Arch Clin Neuropsychol. 2019;34(8):1346-1355. doi:
10.1093/arclin/acy098
PubMedGoogle ScholarCrossref 43.Saxbe
D, Khoddam
H, Piero
LD,
et al. Community violence exposure in early adolescence: longitudinal associations with hippocampal and amygdala volume and resting state connectivity.
Dev Sci. 2018;21(6):e12686. doi:
10.1111/desc.12686
PubMedGoogle Scholar 44.Butler
O, Yang
XF, Laube
C, Kühn
S, Immordino-Yang
MH. Community violence exposure correlates with smaller gray matter volume and lower IQ in urban adolescents.
Hum Brain Mapp. 2018;39(5):2088-2097. doi:
10.1002/hbm.23988
PubMedGoogle ScholarCrossref 47.Small
ML, Feldman
J. Ethnographic evidence, heterogeneity, and neighbourhood effects after moving to opportunity. In: van Ham
M, Manley
D, Bailey
N, Simpson
L, Maclennan
D, eds.
Neighbourhood Effects Research: New Perspectives. Springer Netherlands; 2012:57-77. doi:
10.1007/978-94-007-2309-2_3 48.Small
ML, Manduca
RA, Johnston
WR. Ethnography, neighborhood effects, and the rising heterogeneity of poor neighborhoods across cities.
City Community. 2018;17(3):565-589. doi:
10.1111/cico.12316
Google ScholarCrossref 49.Lloyd
JEV, Hertzman
C. How neighborhoods matter for rural and urban children’s language and cognitive development at kindergarten and grade 4.
J Community Psychol. 2010;38(3):293-313. doi:
10.1002/jcop.20365
Google ScholarCrossref 50.Dea
C, Gauvin
L, Fournier
M, Goldfeld
S. Does place matter? An international comparison of early childhood development outcomes between the metropolitan areas of Melbourne, Australia and Montreal, Canada.
Int J Environ Res Public Health. 2019;16(16):2915. doi:
10.3390/ijerph16162915
PubMedGoogle ScholarCrossref 51.Chase-Lansdale
PL, Gordon
RA. Economic hardship and the development of five- and six-year-olds: neighborhood and regional perspectives.
Child Dev. 1996;67(6):3338-3367. doi:
10.2307/1131782
Google ScholarCrossref 53.Burdick-Will
J, Ludwig
J, Raudenbush
S, Sampson
RJ, Sanbonmatsu
L, Sharkey
P. Converging evidence for neighborhood effects on children’s test scores: an experimental, quasi-experimental, and observational comparison. In: Duncan
GJ, Murnane
RJ, eds. Whither Opportunity? Rising Inequality, Schools, and Children's Life Chances. Russell Sage Foundation; 2011:255-276.
58.Luciana
M, Bjork
JM, Nagel
BJ,
et al. Adolescent neurocognitive development and impacts of substance use: overview of the adolescent brain cognitive development (ABCD) baseline neurocognition battery.
Dev Cogn Neurosci. 2018;32:67-79. doi:
10.1016/j.dcn.2018.02.006
PubMedGoogle ScholarCrossref 62.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
PubMedGoogle ScholarCrossref 65.Diemer
MA, Mistry
RS, Wadsworth
ME, López
I, Reimers
F. Best practices in conceptualizing and measuring social class in psychological research.
Anal Soc Issues Public Policy. 2013;13(1):77-113. doi:
10.1111/asap.12001
Google ScholarCrossref 67.Subramanian
SV, Kubzansky
L, Berkman
L, Fay
M, Kawachi
I. Neighborhood effects on the self-rated health of elders: uncovering the relative importance of structural and service-related neighborhood environments.
J Gerontol B Psychol Sci Soc Sci. 2006;61(3):S153-S160. doi:
10.1093/geronb/61.3.S153
PubMedGoogle ScholarCrossref 69.Gershon
RC, Slotkin
J, Manly
JJ,
et al. IV. NIH Toolbox Cognition Battery (CB): measuring language (vocabulary comprehension and reading decoding).
Monogr Soc Res Child Dev. 2013;78(4):49-69. doi:
10.1111/mono.12034
PubMedGoogle ScholarCrossref 70.Zelazo
PD, Anderson
JE, Richler
J, Wallner-Allen
K, Beaumont
JL, Weintraub
S. II. NIH Toolbox Cognition Battery (CB): measuring executive function and attention.
Monogr Soc Res Child Dev. 2013;78(4):16-33. doi:
10.1111/mono.12032
PubMedGoogle ScholarCrossref 71.Akshoomoff
N, Beaumont
JL, Bauer
PJ,
et al. VIII. NIH Toolbox Cognition Battery (CB): composite scores of crystallized, fluid, and overall cognition.
Monogr Soc Res Child Dev. 2013;78(4):119-132. doi:
10.1111/mono.12038
PubMedGoogle ScholarCrossref 75.Mujahid
MS, Diez Roux
AV, Morenoff
JD, Raghunathan
T. Assessing the measurement properties of neighborhood scales: from psychometrics to ecometrics.
Am J Epidemiol. 2007;165(8):858-867. doi:
10.1093/aje/kwm040
PubMedGoogle ScholarCrossref 77.Berhane
K, Gauderman
WJ, Stram
DO, Thomas
DC. Statistical issues in studies of the long-term effects of air pollution: the Southern California Children’s Health Study.
Stat Sci. 2004;19(3):414-449. doi:
10.1214/088342304000000413
Google ScholarCrossref 82.Wight
RG, Aneshensel
CS, Miller-Martinez
D,
et al. Urban neighborhood context, educational attainment, and cognitive function among older adults.
Am J Epidemiol. 2006;163(12):1071-1078. doi:
10.1093/aje/kwj176
PubMedGoogle ScholarCrossref 83.Scott
AB, Reed
RG, Garcia-Willingham
NE, Lawrence
KA, Segerstrom
SC. Lifespan socioeconomic context: associations with cognitive functioning in later life.
J Gerontol B Psychol Sci Soc Sci. 2019;74(1):113-125. doi:
10.1093/geronb/gby071
PubMedGoogle ScholarCrossref 85.Gianaros
PJ, Kuan
DCH, Marsland
AL,
et al. Community socioeconomic disadvantage in midlife relates to cortical morphology via neuroendocrine and cardiometabolic pathways.
Cereb Cortex. 2017;27(1):460-473. doi:
10.1093/cercor/bhv233PubMedGoogle Scholar 86.Amso
D, Lynn
A. Distinctive mechanisms of adversity and socioeconomic inequality in child development: a review and recommendations for evidence-based policy.
Policy Insights Behav Brain Sci. 2017;4(2):139-146. doi:
10.1177/2372732217721933
PubMedGoogle ScholarCrossref 88.Gard
AM, Maxwell
AM, Shaw
DS,
et al. Beyond family-level adversities: exploring the developmental timing of neighborhood disadvantage effects on the brain.
Dev Sci. 2021;24(1):e12985. doi:
10.1111/desc.12985PubMedGoogle Scholar 90.Gellci
K, Marusak
HA, Peters
C, Elrahal
F, Iadipaolo
AS, Rabinak
CA. Community and household-level socioeconomic disadvantage and functional organization of the salience and emotion network in children and adolescents.
Neuroimage. 2019;184:729-740. doi:
10.1016/j.neuroimage.2018.09.077
PubMedGoogle ScholarCrossref 92.Tamnes
CK, Herting
MM, Goddings
AL,
et al. Development of the cerebral cortex across adolescence: a multisample study of inter-related longitudinal changes in cortical volume, surface area, and thickness.
J Neurosci. 2017;37(12):3402-3412. doi:
10.1523/JNEUROSCI.3302-16.2017
PubMedGoogle ScholarCrossref 98.Staff
RT, Murray
AD, Ahearn
TS, Mustafa
N, Fox
HC, Whalley
LJ. Childhood socioeconomic status and adult brain size: childhood socioeconomic status influences adult hippocampal size.
Ann Neurol. 2012;71(5):653-660. doi:
10.1002/ana.22631
PubMedGoogle ScholarCrossref 103.Hackman
DA, Kraemer
DJM. Socioeconomic disparities in achievement: insights on neurocognitive development and educational interventions. In: Thomas
MSC, Mareschal
D, Dumontheil
I, eds.
Educational Neuroscience: Development Across the Life Span. Frontiers of Developmental Science. Routledge Books; 2020:88-120. doi:
10.4324/9781003016830-6 114.Kessler
RC, Duncan
GJ, Gennetian
LA,
et al. Associations of housing mobility interventions for children in high-poverty neighborhoods with subsequent mental disorders during adolescence.
JAMA. 2014;311(9):937-948. doi:
10.1001/jama.2014.607PubMedGoogle ScholarCrossref 120.Merz
EC, Maskus
EA, Melvin
SA, He
X, Noble
KG. Socioeconomic disparities in language input are associated with children’s language-related brain structure and reading skills.
Child Dev. 2020;91(3):846-860. doi:
10.1111/cdev.13239
PubMedGoogle ScholarCrossref