Exploring networks of complex developmental trauma symptomatology among children and adolescents involved in child welfare

Abstract Background Clinical presentations of child and adolescent psychopathology can vary systematically for boys and girls. While network analysis is increasingly being applied to explore psychopathology in adults, there is a dearth of network studies considering differences in symptoms for boys and girls, particularly in developmental trauma‐related symptomatology. Methods This study involves rural children (n = 375, 39.47% girls) and adolescents (n = 291, 51.20% girls) involved with child protection services in Ontario, Canada. Caregivers completed the Assessment Checklist for Children or Adolescents within the first 6 months of care. Psychometric network analyses were conducted using subscales for boys and girls. Differences were examined via network comparison permutation tests, moderated network models, and independent t‐tests. Results Attachment‐related interpersonal difficulties were the most central nodes in the child and adolescent networks for both boys and girls. Emotional dysregulation also had high strength centrality for adolescents. While network comparison tests found the overall network structures and global network strength to be invariant between boys and girls for children and adolescents, moderated network models and independent t‐tests revealed several differences with regards to the expression of specific symptoms. Among children, girls exhibited more indiscriminate and pseudomature interpersonal behaviors, whereas boys expressed significantly more non‐reciprocal interpersonal behaviors and self‐injury. Adolescent girls exhibited more behavioral dysregulation and suicide discourse in the moderated network model; t‐tests also indicated higher levels of emotional dysregulation, negative self‐image, and other items considered clinically important complex trauma symptoms (e.g., distrust of adults, confused belonging). Discussion This study supports evidence of differences in the expression of complex trauma symptomatology for boys and girls. Additionally, girls exhibit more symptoms, in general. Consistent with the transdiagnostic conceptualization of the consequences of developmental trauma, findings demonstrate the primacy of attachment‐specific difficulties and emotion dysregulation.


INTRODUCTION
Recent perspectives on psychopathology suggest that mental health challenges, including post-traumatic stress disorder (PTSD), may be best conceptualized as a systemic network wherein symptoms are mutually reinforcing and interdependently influential (Borsboom et al., 2021).To date, several studies have examined PTSD symptom networks in children and adolescents (Bartels et al., 2019;De Haan et al., 2020;Scharpf et al., 2022), but none have specifically explored complex trauma symptoms.Moreover, previous work indicates differences between boys and girls in rates and domains of psychopathology across childhood and adolescence, including trauma-related symptoms (Hayward & Sanborn, 2002;Hudziak et al., 2007), yet no studies have examined these gender differences from a network perspective.Thus, the present study furthers extant knowledge on complex trauma by investigating differences in symptom networks for boys and girls among children and adolescents who have experienced maltreatment (i.e., neglect and abuse) and are involved with child welfare in Ontario, Canada.

Complex developmental trauma
The term complex developmental trauma describes both children's exposure to multiple traumatic events and the immediate and longterm effects of this exposure (Cook et al., 2003).In terms of exposure, complex developmental trauma has been defined as the "… experience of multiple, chronic and prolonged, developmentally adverse traumatic experiences, most often of an interpersonal nature (e.g., sexual or physical abuse, war, community violence) and with early-life onset" (van der Kolk, 2005, p. 402).Complex developmental trauma yields a wide range of detrimental immediate and long-term neurodevelopmental and psychosocial impairments, especially when the exposures occur during critical or sensitive developmental periods (Nelson et al., 2019;Nelson & Gabard-Durnam, 2020).Such experiences initiate a negative developmental cascade that affects biological, psychological, emotional, and social processes over time (Cicchetti & Toth, 2016;D'Andrea et al., 2012;Perry, 2009;Sheridan & McLaughlin, 2020;Toth & Manly, 2018).Relatedly, attachmentbased symptomatology seems most prevalent in the early years, while difficulties related to self-esteem, self-concept, substance use, and suicidality become more apparent in adolescence.
Retrospective studies have estimated that the combination of interpersonal victimization and disrupted primary caregiving can explain 45% of the variance in the population attributable risk for childhood-onset psychopathologies (Green et al., 2010;Norman et al., 2012;Teicher & Samson, 2013).Some have even suggested that individuals with psychopathology and who have a history of childhood maltreatment comprise a clinically and neurobiologically distinct subgroup.Teicher and Samson (2013) argue that individuals in this subgroup have earlier onset, greater symptom severity, more comorbidity, more consistent reductions in hippocampal volume and amygdala hyperreactivity, a greater risk for suicide, and poorer treatment outcomes.
Since Herman (1992) first introduced the term complex trauma into the psychiatric lexicon, several studies have attempted to capture the scope of complex developmental trauma sequalae (Cook et al., 2003, van der Kolk, 2005).In these papers, the authors reviewed literature linking complex developmental trauma exposure to several domains of impairment, including attachment disruption, biological and somatic dysregulation, affect regulation, dissociation, issues with behavioral and impulse control, cognitive and attentional impairments, and compromised self-concept.A review by D 'Andrea et al. (2012) subsequently identified symptoms related to affect and behavioral dysregulation, disturbances of attention and consciousness, distortions in attributions, and interpersonal difficulties.In addition, many of the studies reviewed by D'Andrea and colleagues indicated that the expression of symptoms and biopsychosocial impairments tended to occur when children and youth had multiple experiences of interpersonal trauma and should therefore be considered as interrelated symptoms rather than independent, as per the dictates of the current psychiatric nosology.
Children exposed to traumatic experiences exhibit symptom profiles that do not fit internalizing-externalizing dimensions and often do not meet full criteria for PTSD (Ford et al., 2022;Spinazzola Cloitre et al., 2018).
Given the complexity of the sequelae described above, individuals with histories of complex developmental trauma are often diagnosed with multiple comorbid disorders, which are viewed as discrete problems to be addressed rather than as facets of one larger problem (Herman, 1992;D'Andrea et al., 2012;Kretschmar et al., 2017).Such multi-diagnostic formulations pose challenges for children, caregivers, and clinicians when it comes to understanding children's difficulties in the context of their trauma histories, as well as being harder to treat (Nanni et al., 2012;Shenk et al., 2014).This is particularly evident in children involved in child welfare, as they tend to represent the extreme end of the distribution with respect to exposure to abuse and neglect and symptom severity and chronicity (Fisher, 2015;Pace et al., 2019;Tarren-Sweeney, 2008).CPTSD is characterized by three symptom clusters related to disturbances in self-organization (DSO), including negative self-concept, emotional dysregulation, and disturbances in relationships, in addition to the three PTSD criteria (re-experiencing, avoidance, and sense of current threat) (WHO, 2023).While the ICD-11 CPTSD is not specifically a developmental disorder like DTD (i.e., symptom onset can occur in adulthood), several developmental presentations are noted, including a recognition that children and adolescents are more vulnerable than adults to developing CPTSD when exposed to severe or prolonged traumata, neurobiological consequences in children and adolescents can result in cognitive difficulties and executive functioning deficits, distinct behavioral presentations of emotional dysregulation and interpersonal difficulties can be observed among children and adolescents, and that children and adolescents, for whom the source of the trauma was their parent(s) or primary caregiver(s), often develop a disorganized attachment style (WHO, 2023).Factor analytic work has supported the distinction between ICD-11 PTSD and CPTSD in children and adolescents (Haselgruber et al., 2020;Perkonigg et al., 2016;Sachser et al., 2017).

Gender differences in (complex) trauma sequalae
A nuanced understanding of developmental psychopathology necessitates examination of gender-based differences, which are present in general mental health difficulties and trauma-specific symptoms.The focus of this paper concerns differences for boys and girls, though differences in trauma-related symptomatology across other genders is an important area of inquiry.In childhood and adolescence, girls tend to exhibit more internalizing psychopathology symptoms, which may increase the use of maladaptive coping strategies (e.g., rumination), create more conflictual social relationships (Martin & Hadwin, 2022), and result in lower life satisfaction (Campbell et al., 2021;Carragher et al., 2016;Yoon et al., 2022).Conversely, boys tend to exhibit higher externalizing challenges and more disruptive behavior (Carragher et al., 2016).A similar pattern is found in trauma-exposed children and adolescents (Feiring et al., 1999;Gray & Rarick, 2018), though studies with chronically victimized children/youth suggest that girls show higher severity (Ascendio et al., 2022) and a greater variety of symptoms, regardless of the symptom type (Wamser-Nanney & Cherry, 2018), and are more likely to meet proposed CPTSD diagnostic criteria (Hébert & Amédée, 2020;Redican et al., 2022;Sachser et al., 2017;Solva et al., 2020).Behavioral manifestations of difficulties (e.g., aggression) are more apparent among boys, while sexual related trauma symptoms and self-esteem problems are more common amongst girls.With that said, not all studies are consistent in finding gender differences with regards to complex trauma responses (Brewin et al., 2017), and more research is needed to determine factors that may influence differential response patterns, including the expressions of specific types of symptoms.Furthermore, most of the existing research that has considered gender differences has focused on average differences in the expression of specific symptoms and rates of meeting diagnostic criteria; consequently, differences in the interrelationships of trauma-related symptomatology across these groupings remains underexplored.Ongoing research exploring gender-based differences in trauma symptomatology will enable more specific, clientcentered etiological models, informed by unique patterns of prognosis, prevalence, and symptom profiles (Hartung & Lefler, 2019; Child and adolescent trauma symptom networks Network analysis has gained stride as a tool for establishing the empirical foundation of the network perspective of psychopathology, which conceptualizes mental disorders as the product of complex networks of associated symptoms, rather than as the common cause of symptoms (Borsboom, 2017;Fried et al., 2017).The network approach redresses shortcomings of the traditional (common cause) perspective-such as the inability to account for the co-occurrence of symptoms (Borsboom & Cramer, 2013)-by conceptualizing disorders as a system that arises from symptom interactions, studied in their full complexity without imposed boundary restrictions.Cook et al., 2005;D'Andrea et al., 2012;van der Kolk, 2005van der Kolk, , 2009)).Thus, there is a need for network studies to employ measures of complex trauma specifically tailored to such settings when considering symptom profiles among children and adolescents.

Research questions
The present study examines the interrelationships among complex trauma-related symptoms in child and adolescent boys and girls living in rural Ontario, Canada, who are involved with child protection services due to substantiated maltreatment.We explored: (RQ1) How are different symptoms of complex trauma interrelated in children and adolescents who have been exposed to interpersonal trauma, (RQ2) Are there gender-related differences in the structure of complex trauma symptoms amongst children and adolescents, (RQ3) Does the strength centrality of attachment-related symptoms differ in networks amongst boys and girls, and (RQ4) Do children and adolescents show varying patterns of gender-related differences in the structure of complex trauma symptoms?Given that network analysis research is exploratory in nature, we did not have specific hypotheses for the structure of the networks.However, we expected girl networks to be denser (i.e., more significant connections between nodes) and with higher strength of conditional associations between symptoms (i.e., higher global strength) and for attachment-specific and dissociation-related symptoms to be more influential over the network (i.e., higher strength centrality).We also anticipated behavioral dysregulation symptoms to have higher strength centrality in boy networks.
Normative, borderline, and clinical ranges for ACC subscale raw scores can be found in Table S1 in the supplemental file.The ACC shows strong psychometric properties in previous research with children in out-of-home care (⍺ = 0.70-0.89;Tarren-Sweeney, 2014).
Cronbach's ⍺ on each ACC subscale in the present study were mostly acceptable or excellent; however, two subscales (Anxious Distrustful and Sexual) were borderline acceptable (both ⍺ = 0.68) and we excluded the Abnormal Pain Response and Other Items subscales due to poor internal reliability (see Table S1 in the supplemental file).

Assessment Checklist for Adolescents
Adolescents' complex trauma symptoms were assessed using the Assessment Checklist for Adolescents (ACA; Tarren-Sweeney, 2013b).
Totals were derived by summing all items on that scale, and higher scores reflect more challenges.Normative, borderline, and clinical ranges for ACA subscale raw scores can be found in Table S1 in the supplemental file.Previous evaluations of the ACA have shown strong psychometric properties based on adolescents in out-of-home care (⍺ = 0.76-0.92;Tarren-Sweeney, 2014).Cronbach's ⍺ in the present study were acceptable to excellent, aside from the Dissociation and Trauma Symptoms subscale (⍺ = 0.57; see Supplemental Table S2).We chose to retain this scale due to the relevance for the present analyses.

Descriptive statistics and outlier management
All analyses were performed in RStudio Version 4.1.2.We examined the distribution and normality of the ACC and ACA by calculating each subscale's mean, standard deviation, skew, and kurtosis using the psych package (Revelle, 2022).All analyses were conducted with and without outliers being winsorized to examine the extent of outlier influence.Results presented are with the unmodified data due to the winsorized dataset yielding nearly identical results.

Psychometric network analysis
Psychometric network analysis captures the potential inter-related nature of symptoms (Bringmann et al., 2022), which may give rise to a broader disorder profile (e.g., CPTSD) rather than treating symptoms as being caused by a common CPTSD factor.Thus, instead of assuming that there is a generalized CPTSD factor causing negative self-concept, emotion dysregulation, interpersonal difficulties, re-experiencing, avoidance, and a sense of being under threat, a network of these mutually-reinforcing symptoms could jointly yield a clinical presentation of CPTSD (Fried, 2022;Fried et al., 2017).
Taking such an approach can be beneficial for examining meaningful connections among symptoms, lending insight into the nature of how symptoms are interrelated and, consequently, promising treatment targets (Fried et al., 2017;Henry et al., 2022).
We estimated child and adolescent-specific Gaussian graphical models (GGM; i.e., a network of partial correlations using continuous variables) for boys and girls, as differences in the ACC and ACA precluded the merging of the child and adolescent data.We used the bootnet R package (Epskamp et al., 2018) to determine the number of statistically significant edges for all generated networks (p < 0.05), as well as the mean edge weight (i.e., average of edge weights in the network).To estimate networks, we applied partial correlation regularization techniques (i.e., least absolute shrinkage and selection operator [LASSO] regularization with the Extended Bayesian Information Criterion (Foygel & Drton, 2010) with a hyperparameter of 0.5) to identify definitive, parsimonious networks.LASSO regularization is a model selection approach that involves testing the fit of a series of models with various edges constrained to zero, in order to determine the model that fits the data best (Tibshirani, 1996).
As recommended by Epskamp et al. (2018), we examined the accuracy and stability of networks via bootstrapping procedures (2500 draws).This enabled examinations of the accuracy of estimated edge-weights, the stability of node centrality indices, and testing for significant differences between edge-weights and centralities.The stability of centrality indices was quantified via the CScoefficient (Epskamp et al., 2018), which reflects the maximum proportion of cases that can be dropped to retain a correlation with the original centrality of higher than 0.70, with 95% certainty.CScoefficients above 0.25 were interpreted (Epskamp et al., 2018).
With regards to centrality metrics, we focused on node strength (i.e., how influential a symptom is in the network, calculated by summing the absolute edge weights of edges per symptom; see RQ2) as it is of particular relevance for psychological networks (Bringmann et al., 2019), though other common centrality measures, including closeness, betweenness, and expected influence, are reported in the supplemental materials.We used the qgraph (Epskamp et al., 2012) package for network visualizations, which are comprised of nodes (signifying symptoms) and edges (signifying the conditional relationship-partial correlationsbetween two symptoms).In the GGMs, the distance between nodes as well as edge width and transparency reflect the strength of the association between nodes.Thus, nodes closer together with wider and darker edges have stronger conditional associations.Missing edges indicate a non-significant partial correlation.Blue and red edges indicate positive and negative associations, respectively.

Network comparison test
We compared networks for boys and girls with the NetworkCompar-isonTest package (Van Borkulo et al., 2023).This permutation test

Symptom severity comparison
In addition to the NCT, which compares network structures and overall network strength, we examined differences in associations between boys and girls with each symptom by including a dummy variable for gender (boy = 0, girl = 1) as a node in moderated network models (Haslbeck, 2022) for child and adolescent symptoms using the bootnet package.To examine the degree to which moderation effects emerge with and without controlling for all other effects, we also conducted independent samples t-tests for all reliable ACC and ACA subscales using the stats package (R Core Team, 2021).
Tests for homogeneity of variance were conducted before computing t-tests.Student's t-tests were conducted for comparisons in which the variances were found to be homogenous; Welch's t-tests were conducted in cases of heteroscedasticity.We corrected for multiple comparisons using the false discovery rate control method (Glickman et al., 2014) to minimize type I error.3).Children had an average of M = 3.02 (SD = 2.05) and adolescents had an average of M = 2.84 (SD = 1.84) clinical scales above the clinical cut-off.

Network structure
The networks of interpersonal trauma symptoms based on the ACC Edge weights in the girl network ranged from −0.10 (sexual behaviornegative self-image) to 0.44 (negative self-image-low confidence), with a mean of 0.09.Weight matrices for all networks can be found in the supplementary materials (Tables S3 and S4).

Node strength
In the child networks, the most central complex trauma symptoms for  CIs.The CS-coefficient for strength centrality for the four networks was 0.59, 0.36, 0.51 and 0.60 for boys and girls in the child and adolescent networks, respectively, exceeding the recommended thresholds for stable estimation (Epskamp et al., 2018).Overall, these metrics indicate that the stability of both edge-weights and strength centrality indices in the estimated networks were acceptable.Additional details are available in the supplementary materials, including in Table S6.

Network comparison test
The network comparison test for the child sample revealed that the global strength and structure of the networks for the boys and girls were invariant (S = 0.14, p = 0.788; M = 0.20, p = 0.440).Similarly, the networks for adolescent boys and girls were invariant (S = 0.08, p = 0.772; M = 0.19, p = 0.645).Due to the invariance in the network structures (i.e., the M statistic), we did not proceed with testing invariance in the strength of specific edges in either the child or adolescent samples.

Pairwise comparisons
The moderated network model in the child sample (see Figure 3 The magnitude of effect sizes for pairwise comparisons, as measured by Cohen's d, were small.See Table 2 for respective scale means and effect sizes. In the adolescent sample, gender significantly interacted with behavioral dysregulation and suicide discourse, such that girls experienced significantly higher levels of both, after controlling for all other effects (Figure 3, right panel and adolescents in out-of-home care (Denton et al., 2017).Hence, we aimed to enhance the current understanding of gender differences in complex trauma-related symptomology and their interrelations through the application of psychometric network analysis.

Network composition of complex trauma symptomatology
The centrality indices in all networks underscore the primacy of the attachment-specific symptoms, negative self-concept, and emotional dysregulation, irrespective of developmental stage and gender.The  (Bartels et al., 2019;De Haan et al., 2020;Scharpf et al., 2022).Together with the importance of attachment dysregulation symptoms, these patterns support the transdiagnostic model of developmental mechanisms that explain the link between childhood trauma and psychopathology (McLaughlin et al., 2020) and is consistent with the conceptualization of developmental trauma disorder (DTD; van der Kolk et al., 2009).

Gender differences in complex trauma symptomatology
This study found that the network structure of complex trauma symptomatology in children and adolescents, as measured by the ACC and ACA, respectively, were invariant across boys and girls.This indicates that interrelationships among complex trauma symptoms are consistent in boys and girls in both childhood and adolescence, lending credence to the conceptualization of complex developmental trauma as a unitary construct that is appropriate for boys and girls.
Boys, on the other hand, were found to exhibit more behaviors indicative of dismissive-avoidant attachment (emotionally withdrawn, avoidant, and not reciprocating in social and emotional exchanges) and emotional dysregulation resulting in self-harming behaviors.The nonreciprocal interpersonal behaviors exhibited by boys, if not conceptualized from an attachment theory perspective, may be erroneously diagnosed as oppositional defiance and as an early indication of antisocial traits (Ford et al., 2022).Such an interpersonal style may not only arise from violent victimization-which occurs more frequently in boys and can be related to higher levels of surgency and hyperactivity-but may be interpreted as intentional displays of opposition, contributing to future physical violence in high-risk contexts (Fallon et al., 2022;Finkelhor et al., 2015;Moody et al., 2018).

Symptom-specific gender differences in adolescents
We found adolescent girls to exhibit higher dysregulation symptomology, both emotionally and behaviorally.This contrasts the widely replicated finding, from studies with different measurement protocols and samples, that girls display more internalizing challenges and boys higher externalizing difficulties (Carragher et al., 2016), highlighting the different developmental trajectories adolescents may undergo after developmental trauma experiences.In tandem, the symptom profiles of adolescent girls are more elevated across the board in comparison to boys, including having more negative selfconcepts and higher rates of suicidality.Trauma experiences can have diverse, extensive impacts, and adolescent girls may be especially vulnerable to their effects.Additionally, girls are more likely to engage in maladaptive coping strategies such as rumination, which can exacerbate negatively valenced emotions and thoughts that stem from complex trauma (Martin & Hadwin, 2022).Conversely, boys may engage in alternative coping mechanisms, such as substance use, or use avoidance-based strategies that suppress their experiences, leading to lower symptom reports by caregivers.

Limitations and implications
The current study included several limitations.The analysis of gender differences was limited to boys and girls, and therefore did not include potential differential symptom expressions in gender-diverse children and adolescents, due to pre-existing limitations within the existing data collected by TFCP.That said, the rate of non-binary gender identities with substantiated cases of maltreatment in the Ontario child welfare system is very low (<1%; Fallon et al., 2022), which would have precluded analyses within the present sample if such data were available.Other key sociodemographic data were unavailable in the study, including types and chronicity of trauma exposure, socioeconomic status, and racial/ethnic identity.Differential rates of exposure to types of maltreatment among boys and girls have been found (Fallon et al., 2022;Wamser-Nanney & Cherry, 2018), which suggests a need for more specific gender-based analyses accounting for trauma exposure.For example, Wamser-Nanney and Cherry (2018) found gender-differences in symptoms to persist among children exposed to child sexual abuse.Gender differences in symptom networks for specific types of maltreatment (e.g., neglect, physical, emotional, and sexual abuse) will be an important future direction.The present sample comes from rural catchment areas with predominantly white families.Also, ethnicity/ race data were not systematically collected by the partnering agency.
Hence, exploring the expression of complex trauma symptoms in diverse groups of children and adolescents is an important future direction.Additional moderators of interests, including care type and age into care, are also important considerations for future research examining complex trauma symptoms networks in maltreated children and youth.
The single-informant methodology was another limitation, as the caregivers were the only source of information.In the future, it will be beneficial to collect data from various individuals who know the child/adolescent so that their symptoms can be evaluated from

Implications and conclusion
Despite the above limitations, several implications for clinical practice with children and adolescents who have experienced developmental trauma, and who are involved in the child welfare system, emerge from our results.The findings underscore the primacy of attachment insecurity and emotional dysregulation as being a central mechanism relevant to children's and adolescents' complex trauma symptomatology, consistent with the emergent conceptualization of complex developmental trauma (Cook et al., 2005;D'Andrea et al., 2012;McLaughlin et al., 2020;van der Kolk, 2005).Clinical assessment and practice should, therefore, intentionally consider the administration of validated, developmentally appropriate measures of complex trauma symptomatology that includes measures of attachment (such as the ACC/ACA).Further, the centrality of attachment-specific symptoms and emotional dysregulation indicate that interventions that intentionally target improving attachment security and emotion regulation capacities may be particularly potent (Fried et al., 2017;Henry et al., 2022).This is consistent with a previous longitudinal study with the present child sample that found improvements in NETWORKS OF COMPLEX DEVELOPMENTAL TRAUMA SYMPTOMS

��
Previous research indicates differential expressions of complex trauma among boys and girls, but limited work has examined if the interrelationship between symptoms differ.Across gender and developmental stage, attachmentrelated difficulties were most central.� Consistent with previous research, the present study observed gender differences in the presentation of complex trauma symptomatology and found that girls, regardless of developmental stage, tend to exhibit more symptoms than boys.� The structure of complex trauma symptom networks is invariant between boys and girls in childhood and adolescence.� Clinical assessment and treatment should use validated, developmentally appropriate measures of complex trauma symptomatology, which include measures of attachment security and emotional dysregulation.Further, the centrality of attachment-specific symptoms and emotional dysregulation indicate that interventions that target attachment security and emotion regulation may be particularly potent.al., 2018).Relatedly, common measures of trauma for children and adolescents fail to capture the unique symptoms expressed by children who have experienced maltreatment, including those involved in out-of-home care and child protection services (Denton et al., 2017; Tarren-Sweeney, 2013a).However, several questionnaires, the majority of which are caregiver-report measures, have been developed to capture trauma sequalae beyond PTSD symptoms (see Denton et al., 2017 for a review), including the caregiver-report Assessment Checklist for Children (ACC; Tarren-Sweeney, 2007) and Adolescents (ACA; Tarren-Sweeney, 2013b) and the self-report International Trauma Questionnaire for Children and Adolescents (ITQ-CA; trauma's unique neurobiological and phenotypic signature, the National Child Traumatic Stress Network developed criteria for a new disorder, Developmental Trauma Disorder (DTD), proposed to be included in the fifth edition of the American Psychiatric Association's (APA) Diagnostic and Statistical Manual of Mental Disorders (DSM-5).The proposed diagnostic criteria included the following: Lifetime contemporaneous exposure to interpersonal victimization and primary caregiver attachment disruption (Criterion A); Affective and Somatic Dysregulation (Criterion B); Attentional and Behavioral Dysregulation (Criterion C); and Self and Relational Dysregulation (Criterion D) (see Ford et al., 2018 for the full criteria).A systematic review by Morelli and Villodas (2021) examined the construct validity and clinical utility of DTD based on the 21 empirical studies (including 17 separate samples) published between 2005 and July 2021.The studies largely supported the DTD construct, including divergent and convergent validity and complex trauma exposure being associated with DTD symptoms.Additionally, factor analytic studies consistently found a three-factor solution for DTD symptoms measured with a variety of DTD-specific and ad hoc DTD scales to offer the best fit to the data.However, the review highlights the need for more research with larger samples by research groups who are not associated with the developers of the construct and for the use of more rigorous methodological and analytic designs.Parallel efforts to those for the DSM, which have aimed to refine the diagnostic precision and clinical utility of the World Health Organization's (WHO) International Classification of Diseases (ICD) classification of trauma responses, led to the formal inclusion of complex PTSD (CPTSD) in the 11th version of the ICD (WHO, 2018).
Several recent studies have examined networks of PTSD and comorbid symptoms in children and youth who have experienced trauma and adversity.Using self-and caregiver-report data from an international clinical sample of children and adolescents, Bartels et al. (2019) identified negative cognitions and emotion dysregulation as the most central nodes in self-report symptom networks; challenges related to re-experiencing and hypervigilance were additionally central in the caregiver-report network.De Haan et al. (2020) also investigated PTSD and depression symptoms in an international sample of children and youth, finding re-experiencing-related PTSD symptoms and impaired concentration in depression to be the most central.Additionally, Scharpf et al. (2022) examined PTSD symptom networks in a sample of war-affected children and adolescents from across six countries.Re-experiencing and avoidance were most strongly connected to other symptoms in the full sample, as well as child-and adolescent-specific networks.These findings collectively illustrate the advantages of network analysis for understanding trauma-related symptoms.However, the studies relied on PTSD diagnostic criteria, which fail to capture the full breadth of symptoms expressed by children and adolescents who have experienced complex developmental trauma (repeated and/or multiple exposures to interpersonal violence and attachment disruptions; Data used in the present study were collected by the Therapeutic Family Care Program (TFCP).TFCP provides therapeutic services to children and adolescents with substantiated cases of maltreatment and their caregivers (biological, kinship, adoptive, and foster caregivers) referred by the Children's Aid Societies of Durham, Kawartha-Haliburton, and Highland Shores in Ontario, Canada.These catchment areas are large, rural regions falling outside nearby urban centers.Biological or alternative caregivers consented to participate in the agency's services, which included routine measures of children's/youth's psychosocial functioning to monitor treatment progress.Children and adolescents expressed verbal assent for participation.This cross-sectional analysis includes data from the first assessment for each child/youth, which were collected between 2000 and 2019.Data were available for 555 children, though we included a subset aged 5-11 years (n = 375; 39.47% girls; M age = 7.6, SD age = 2.1) to match the sample used to validate the assessment measure.Data were available for 291 adolescents (51.20% girls; statistically compares networks on (i) overall network structure (test statistic = M; represents the maximum difference in edge strength of the two networks) and (ii) global network strength (test statistic = S;represents the summed value of all the edges in a network)(Van Borkulo et al., 2023).Significant differences in edge strengths between gender networks can be tested post-hoc for non-invariant network structures.
for boys and girls are visualized in Figure 1.Each network included 11 nodes representing scores on each clinical scale.Networks were moderately sparse: of the 55 possible connections, 34 were retained in the boy network (Ρdensity = 0.62) and 30 were retained in the girl network (Ρdensity = 0.55).Mean edge weights (i.e., partial correlation coefficients) for boys and girls were 0.074 and 0.070, respectively.Edge weights ranged from −0.11 (pseudomature interpersonal behavior -low confidence) to 0.40 (insecure interpersonal behavior-anxiousdistrustful, insecure interpersonal behavior-negative self-image) in the boy network, and from 0.01 (non-reciprocal interpersonal behavior-anxious-distrustful, food maintenance-negative self-image) to 0.36 (pseudomature interpersonal behaviors-non-reciprocal interpersonal behaviors) in the girl network.For adolescents, ACA networks (Figure2) included 10 nodes with 45 possible connections.Of these edges, 26 were retained in the boy network (Ρdensity = 0.58) and 29 were retained in the girl network (Ρdensity = 0.64), suggesting moderate parsimony.For boys, edge weights ranged from 0.02 (sexual behavior-other items) to 0.37 (behavioral dysregulation-emotion dysregulation), with a mean of 0.09.

F I G U R E 1
Network of ACC subscales for children: Boys and girls.ACC1: Sexual behaviors.ACC2: Pseudomature interpersonal behaviors.ACC3: Non-reciprocal interpersonal behaviors.ACC4: Indiscriminate interpersonal behaviors.ACC5: Insecure interpersonal behaviors.ACC6: Anxious-distrustful. ACC7: Food maintenance.ACC8: Self-injury.ACC9: Suicide discourse.ACC10: Negative self-image.ACC11: Low confidence.are reliably two of the strongest edges in the boy network due to their bootstrapped CIs not overlapping with that of several other edges in the network.Additionally, the edge negative self-image-low confidence (ACA9-ACA10) in the network of adolescent girls was reliably stronger than several other edges due to non-overlapping , left panel) revealed that girls expressed significantly more symptoms of indiscriminate and pseudomature interpersonal behaviors, whereas boys expressed significantly more symptoms of non-reciprocal interpersonal behaviors and self-injury.T-tests in the child data were consistent with the moderated network model (indiscriminate interpersonal behaviors: t (373) = −2.88,p = 0.011; pseudomature interpersonal behaviors: t (373) = −3.09,p = 0.011; non-reciprocal interpersonal behaviors: t (355.09)= 3.89, p < 0.001; self-injury: t (369.08)= 3.02, p = 0.011).
ACC included four attachment-specific measures (indiscriminate, insecure, non-reciprocal, and pseudomature interpersonal behaviors) in children.These symptoms had the highest strength centrality in the networks for both boys and girls-especially insecure interpersonal behaviors.Additionally, symptoms related to negative self-identity also emerged as central.While the ACA does not have specific attachment-related subscales, many of the items in the Other Items scale are the same as those from the attachment-specific scales of the ACC.Consistent with the child networks, this subscale had the highest strength centrality for adolescents, in addition to emotion dysregulation, behavioral dysregulation, and those related to negative self-identity (low confidence and negative self-image).The centrality of the emotion dysregulation and negative self-concept symptoms is consistent with previous network studies of PTSD symptoms in children and adolescents multiple perspectives and across various contexts.Given that the measures used in the present study were designed to capture the breadth of symptoms exhibited by children in out-of-home care and predated the proposed diagnostic criteria for Developmental Trauma Disorder and the ICD's Complex PTSD diagnosis, the present study did not examine classic PTSD symptoms, nor did it strictly align with recent frameworks of complex (developmental) trauma.Furthermore, having relatively high-density regularized networks with few nodes (such as in the present study) increases power to detect differences with the network comparison permutation test(van Borkulo et al., 2023).However, estimating the networks with larger samples and more balanced group sizes will also be important in future replication studies, as simulation studies have shown this to increase power to detect differences in the global strength and network structure omnibus tests(van Borkulo et al., 2023).Finally, the cross-sectional design did not allow for an analysis of causality and change over time, and future longitudinal studies should investigate the causes of specific symptom patterns in children and adolescents.
T A B L E 1 a Age Into Care was provided for n = 199 (53.7%) children and n = 138 (47.42%) adolescents.NETWORKS OF COMPLEX DEVELOPMENTAL TRAUMA SYMPTOMS

Table 2
presents the descriptive statistics of the ACC and ACA subscales for children and adolescents, respectively.Based on clinical cut-offs for each of the subscales, and consistent with this being a clinical sample, most children and adolescents had clinically relevant symptoms, with 86.10% of children and 87.30% of adolescents having at least one clinical scale in the clinical range (Table Full sample and gender-specific descriptive statistics of clinical subscales for children and adolescents.Range frequencies and percentages of assessment checklist for children (ACC) and assessment checklist for adolescents (ACA) clinical subscales.
boys, as indicated by standardized strength estimates, were insecure interpersonal behaviors (2.31), non-reciprocal interpersonal behaviors (0.90), negative self-image (0.32), pseudomature interpersonal behaviors (0.31), and indiscriminate interpersonal behaviors (0.06), respectively; suicide discourse, with a value of −1.30, was the least central symptom.For girls, the most central symptoms were insecure interpersonal behaviors (2.04), non-reciprocal interpersonal behaviors (0.59), pseudomature interpersonal behaviors (0.53), low confidence (0.50), indiscriminate interpersonal behaviors (0.47), and negative self-Stability analyses indicated that all four networks were accurately estimated, with moderate confidence intervals (CIs) around the edge T A B L E 2 Note: p-values for Students and Welches t-tests: *p < 0.05, **p < 0.01, ***p < 0.001.All p-values were corrected for multiple comparisons using the False Discovery Rate correction.NETWORKS OF COMPLEX DEVELOPMENTAL TRAUMA SYMPTOMS weights (Figures S3 and S4 in the Supplemental Materials).Most CIs overlapped, implying that edges within networks were unlikely to be significantly different from each other; hence, the order of edge weights should be interpreted with caution.Conversely, the edges for insecure interpersonal behaviors-negative self-image (ACC5-ACC10) and insecure interpersonal behaviors-anxious-distrustful (ACC5-ACC6) T A B L E 3 Network of ACA subscales for children: Boys and girls.ACA1: Non-reciprocal.ACA2: Social instability/behavioral dysregulation.ACA3: Emotional dysregulation/distorted social cognition.ACA4: Dissociation/trauma symptoms.ACA5: Food maintenance.ACA6: Sexual behavior.ACA7: Suicide discourse.ACA8: Other items.ACA9: Negative self-image.ACA10: Low confidence.
-SMITH ET AL.