Heterogeneity in caregiving-related early adversity: Creating stable dimensions and subtypes

Abstract

Early psychosocial adversities exist at many levels, including caregiving-related, extrafamilial, and sociodemographic, which despite their high interrelatedness may have unique impacts on development. In this paper, we focus on caregiving-related early adversities (crEAs) and parse the heterogeneity of crEAs via data reduction techniques that identify experiential cooccurrences. Using network science, we characterized crEA cooccurrences to represent the comorbidity of crEA experiences across a sample of school-age children (n = 258; 6–12 years old) with a history of crEAs. crEA dimensions (variable level) and crEA subtypes (subject level) were identified using parallel factor analysis/principal component analysis and graph-based Louvain community detection. Bagging enhancement with cross-validation provided estimates of robustness. These data-driven dimensions/subtypes showed evidence of stability, transcended traditional sociolegally defined groups, were more homogenous than sociolegally defined groups, and reduced statistical correlations with sociodemographic factors. Finally, random forests showed both unique and common predictive importance of the crEA dimensions/subtypes for childhood mental health symptoms and academic skills. These data-driven outcomes provide additional tools and recommendations for crEA data reduction to inform precision medicine efforts in this area.

Validation of an electronic self-administered version of the Dimensional Inventory of Stress and Trauma Across the Lifespan in a large sample of young adults.

“…However, it is likely that researchers will be increasingly interested in using these highly detailed data to generate dimensions that reflect interactions between specific features of stress. These approaches are likely to require complex multivariate and data-driven approaches (e.g., machine learning-based approaches) that will afford analysis of age- and developmental time period-specific analyses (e.g., Herzog et al, 2020; Nikolaidis et al, 2022; Sicorello et al, 2021). In this vein, depending on the specific research question of interest, we encourage researchers to modify the current “template” of the DISTAL-E to facilitate the assessment of varied additional dimensions of interest that may be pertinent to a diverse array of research questions.…”

Harmonizing bifactor models of psychopathology between distinct assessment instruments: Reliability, measurement invariance, and authenticity

“…A complete description of the RBC samples can be found elsewhere (Hoffmann, Moore, Axelrud, Tottenham, Zuo, et al, 2022). Briefly, it contains phenotypic data from six large-scale developmental imaging cohorts, which are the Philadelphia Neurodevelopmental Cohort (PNC) (Satterthwaite et al, 2016), the Brazilian High-Risk Cohort Study for Mental Conditions (BHRCS) (Salum et al, 2015), the Healthy Brain Network (HBN) (Alexander et al, 2017), the Nathan Kline Institute-Rockland Sample (NKI-RS) (Nooner et al, 2012), the developmental component of the Chinese Color Nest Project (devCCNP) (Liu et al, 2020), and the Parents and Children Coming Together project (PACCT; PIs: Tottenham & Milham) (Nikolaidis et al, 2022). Healthy Brain Network is treatmentseeking samples.…”

Leveraging the developmental neuroscience of caregiving to promote resilience among youth exposed to adversity

“…3). For example, interventions can be tailored based on an individual’s profile of exposure to adversity across multiple dimensions (Cohodes et al, 2021; Cohodes et al, 2023; Nikolaidis et al, 2022), based on specific patterns of caregiver–child interactions (Kitt et al, 2022), or based on specific caregiver-level factors such as parental symptomatology following dyadic exposure to adversity (Hagan et al, 2017). We specifically highlight the importance of considering family-level processes in treatment selection and optimization given the multifaceted influence of caregivers on resilience-related processes reviewed here (Garner et al, 2021).…”