Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2019 Aug 9.
Published in final edited form as: J Juv Justice. 2016 Fall;5(2):68–84.

Childhood Adversity among Court-Involved Youth: Heterogeneous Needs for Prevention and Treatment

Patricia Logan-Greene 1, BK Elizabeth Kim 2, Paula S Nurius 3
PMCID: PMC6688767  NIHMSID: NIHMS1043447  PMID: 31404461

Abstract

Although experiences of trauma and adversity are highly prevalent among juvenile justice-involved youth, few studies examine the heterogeneity of these histories across individuals. This study seeks to inform practitioners regarding the distinct patterns of adversity among this vulnerable population, using an expanded measure of adverse childhood experiences (ACEs). We employed Latent Class Analysis to test for meaningful subgroups of youth based on histories of childhood adversity. The sample (N=5,378) consisted of youth on probation in a western United States county. The best-fitting model contained six classes, described as: Low All (40.3%), Parental Substance Use and Incarceration (12.0%), Poverty and Parental Health Problems (13.2%), High Family Conflict and SES (15.3%), High Maltreatment and Placements (11.0%), and High All (8.1%). Additional testing revealed significant differences across classes in terms of age, gender, race/ethnicity, and living situations. Results strongly support the need to incorporate a trauma-informed framework for both juvenile justice and community service settings, as well as tailoring interventions to meet heterogeneous needs of court-involved youth. Striking variation in the forms and levels of childhood adversity argue the value of screening for ACEs in conjunction with poverty, and working to interrupt problematic trajectories in adolescence and the transition to adulthood.

There is a robust literature examining the overlap of juvenile delinquency with a range of childhood adversities, such as childhood maltreatment, socioeconomic disadvantage, and family dysfunction, including involvement in the child welfare system. These examinations have supported the notion that the majority of youth involved with the juvenile justice systems bring histories of childhood trauma and adversity (Dierkhising, Ko, & Goldman, 2013; Greenwald, 2014). This has led to a growing recognition of the need to transform juvenile justice systems to appropriately address these histories. However, there is as yet little guidance about the specific and heterogeneous needs of court-involved youth with respect to these backgrounds. The present study seeks to fill that gap by testing for distinct patterns of adverse childhood experiences among subgroups of youth involved with the juvenile justice system. The findings of this study can provide practitioners with novel insights regarding distinct adversity profiles with which court-involved youth enter the system, illumining differing patterns of treatment needs.

Childhood Adversity & Court-Involved Youth

One increasingly common way to assess childhood adversity is with the adverse childhood experiences (ACEs) framework. ACEs describe a set of commonly experienced adversities that can be easily assessed in clinical, community, or court settings (Felitti et al., 1998). This builds on a cumulative adversity model wherein exposure to greater numbers of adversities tends to commensurately increase health risks and maladaptive development, especially because negative experiences tend to be interrelated (Anda, Butchart, Felitti, & Brown, 2010; Duke, Pettingell, McMorris, & Borowsky, 2010). ACEs have been found to be interrelated in both broad-based (Dong et al., 2004) and predominantly young minority community samples (Mersky, Topitzes, & Reynolds, 2013) and among court-involved youth (Baglivio & Epps, 2015). Baglivio and Epps (2015) demonstrated that having a single ACE increased the likelihood of having another up to 1,286 times, bolstering the idea that exposures generally do not occur in isolation. Thus, a cumulative assessment better captures the stress load that children’s life contexts impose through which subsequent neurobiological as well as psychosocial pathways can lead to problematic development that cascades across the life course (Logan-Greene, Green, Nurius, & Longhi, 2014; Putnam, 2006).

ACEs assessment has commonly included maltreatment (sexual, physical, and emotional victimization and exposure to family violence and neglect), and family dysfunction (household substance abuse, household illness, incarcerated family member, and parental divorce). When measured as a count of how many adversities an individual has experienced, the ACE score has been shown to be a powerful predictor of health, behaviors, and even morbidity across a wide variety of populations and contexts (Anda, Felitti, Bremner, Walker, Whitfield, & Perry, 2006; Larkin, Shields, & Anda, 2012;Nurius, Green, Logan-Greene, & Borja, 2015). Recent extensions of ACEs have incorporated other indicators of adversity, such as out of home placement in foster care (Cronholm et al., 2015) and family member illness (Wade, Shea, Rubin, & Wood, 2014) among others.

Lacking from the ACEs framework, however, has been an assessment of poverty-related forms of social disadvantage. The impact of socioeconomic disadvantage on health and a range of behavioral outcomes is well established (DeNavas-Walt, 2010; Skowyra & Cocozza, 2007), and may also be entangled with other forms of adversity, such as parental incarceration or illness. Recent work has argued for expanded assessment to include adversities such as poverty, out of home placement, and community threats (Cronholm et al., 2015; Wade et al., 2014) that might further disadvantage young people through the life course. Community based surveys, such as the National Survey of Children’s Health (NSCH), include poverty among the conventional ACEs list (Sacks, Murpheym & Moore, 2014). Recently, Baglivio, Wolff, Epps, and Nelson (2015) found that neighborhood context predicted ACE scores among delinquent youth, such that youth in impoverished census tracts had significantly more ACEs than those in affluent neighborhoods. Thus, this paper adds family socioeconomic disadvantage within an ACEs framework to assess its value in distinguishing household contexts that are posing greater challenges for some system-involved youth than others.

ACE exposures have shown to be particularly high amongst juvenile offenders (Baglivio, Epps, Swartz, Sayedul Hug, Sheer, & Hardt, 2014; Dierkhising et al., 2013; Grevstad, 2010), who also report higher likelihood of experiencing multiple forms of adversity compared to the general population (Abram et al., 2004). A recent study reported that for every additional count of ACEs, the odds of youth becoming a serious, violent, or chronic offender increases by 35%, controlling for other factors (Fox, Perez, Cass, Baglivio, & Epps, 2015). Furthermore, greater ACEs increase the risk of re-arrests, with higher cumulative exposure leading to increased recidivism rates (Wolff, Baglivio, & Piquero, 2015). The findings of these studies suggest a critical need for trauma-informed treatment and services for juvenile-justice involved youth, specific to their complex trauma histories

Testing for Differences in Youth ACE Profiles

This body of work indicates that risks stemming from adversity appear greatest for youth who are nested within contexts that include broader spectrum forms of adversity such as poverty, maltreatment, and parental dysfunction. Variable oriented analytic approaches (such as logistic or linear regression) used in prior work to examine prevalence or linear trends amongst these domains within samples have provided replicated demonstration of step-dose forms of association of cumulative adversity and subsequent health and functioning outcomes. These tools are helpful in characterizing populations overall and have provided a strong foundation as well as an impetus for subsequent stage investigations that test for variation within populations, providing particularly important distinctions within high-risk populations.

Person-oriented analytic methods, such as latent class analysis (LCA), are suited for these latter kinds of questions. Berzenski and Yates (2011), for example, used LCA to ascertain distinct patterns and combinations of four maltreatment experiences. Among students exposed to multiple maltreatment experiences, clusters distinguished homes that were physically violent, those that were emotionally hostile, those that included harsh parenting, and homes that were sexually abusive. Similarly, Mulder and colleagues (2012) used LCA to identify subgroups of offenders (e.g., sex offending group, violent offending group, property offending group) and found that these groups had distinctly different risk profiles leading to differential prediction of recidivism rates. The researchers in the study concluded that these distinct groups and risk profiles indicate need for individualized treatment aiming different risk factors. Additionally, Lanza and Rhodes (2013) argued that LCA is an efficient approach to identify subgroups based on multiple contextual risks and match individualized prevention and treatment needs.

Thus, examining the heterogeneity of the adverse experiences within populations can uncover potentially distinct developmental contexts within which youth are being reared and, thus, differing individual and family treatment and support needs these youth and families have. Person-oriented analytic findings do not stand in opposition to trends established through full sample (or variable-oriented) examination, but rather address complementary questions—such as predicting a phenomenon at a population level relative to aiming to more fully understand variation in developmental mechanisms or pathways—together providing a “binocular view” of the phenomenon in question (Bergman & Trost, 2006, p. 629).

Advances in identifying mechanisms through which early life experiences and environmental influences leave lasting signatures on youth development has emphasized attention to these ecologies of childhood – the social and physical environments within which they are raised (Shonkoff et al., 2012). Given that higher ACE scores indicate a higher risk for impaired developmental trajectories, including early onset and chronic delinquency (Baglivio et al., 2015), it is imperative to ask more penetrating questions regarding differential combinations that these adversities may manifest across system-involved youth. These differences could provide guidance at entrance into the system as to different kinds of programs and services likely to provide stronger prevention and remedial effects. This study is among the first to test for clustering among ACEs within this population. These subgroup-seeking findings provide insights, complementary to studies that average across whole samples, which can guide clinicians in contact with youth in juvenile justice systems in more targeted trauma-informed care (Ford, Chapman, Hawke, & Albert, 2007).

The Present Study

In this paper, we test for empirically supported clustering to determine subsets of court-involved youth who are more like one another than they are the sample as a whole, relative to histories of adverse experiences. We hypothesize that significant clustering will be found, reflecting differing forms of adversity exposure rather than differences in level alone (e.g., low, medium, high), suggesting a strong need for specific treatment approaches. We theorize that these clusters will demonstrate that some traumas may tend to co-occur, providing further detail to variable-oriented framing of cumulative trauma. Additionally, we expect that these clusters will reflect strengths that youth may have, such as a lack of social disadvantage, that service providers may be able to draw upon for interventions. To determine these clusters, we employed Latent Class Analysis (LCA), which is a powerful statistical method used to determine groups of similar individuals within a heterogeneous sample (McCutcheon, 1987). This structure-seeking approach does not have a priori expectations of group compositions yet provides an accurate and complex empirical tool to discern group structure.

Methods

Data

The data come from the Washington State Juvenile Court Assessment (WSJCA) youth adjudicated to probation in an mixed urban and rural, diverse, Western region from 2003 to 2013 (Barnoski, 2004). The assessment tool is called the Positive Achievement Change Tool (PACT), which has been found be valid and empirically sound across gender and racial/ethnic groups (Baglivio & Jackowski, 2013; Barnoski, 2004a; Washington State Institute for Public Policy, 2004). The WSJCA was developed as a two-stage process. In the first-stage, pre-screen assessment was completed for all youth placed on probation to identify low-, moderate-, or high-risk for recidivism. In the second-stage, those identified as moderate- to high-risk for recidivism are given the full assessment that provides a longer and more comprehensive risk and protective factor profile of youth.

Juvenile Probation Counselors (JPCs) in Washington State have been trained to conduct one-on-one interview with youth entering probation and complete the assessment. To further enhance the validity of the assessment, where available, JPCs confirm self-reported responses by contacting other agencies, records (e.g., CPS records) or collateral resources (e.g., parents, teachers, mental health counselors). Assessments completed by probation officers have been found to have acceptable reliability, especially with good training (Barnoski, 2004a).

The sample population includes youth who were identified as moderate- to high-risk during pre-screen assessment based on social and criminal history and had received a minimum of three months’ probation between January 2003 and December 2013. The first case from each youth was included in this analysis, yield a final sample of 5,378 individuals (Female = 23.6%). The average age of the sample youth is 15.5 years, ranging from 10 to 18 years. The racial/ethnic composition of the sample includes 56.0% Caucasian, 24.2% African American, 3.0% American Indian/Alaskan Native, 2.9% Asian American, 1.5% Native Hawaiian or Pacific Islanders, 5.7% Latino, and 6.7% missing, mixed, or other. In this assessment “Latino” is not listed as a different ethnicity category separate from race.

Measures

Demographics.

Youth demographic information regarding age, gender, and race/ethnicity was collected separately as part of the usual system processing. Race was collapsed into four groups, including Caucasian, African American, Latino, and “other,” which included Asian, Hawaiian, Pacific Islander, Native American, and mixed race. Within the WSJCA assessment, questions were asked about youth living situation, allowing youth to endorse any item(s) that reflected their current household composition. Four mutually exclusive variable accessed whether they lived with a biological parent (both biological parent, biological mother only, biological father only, and neither biological parent); another variable assessed whether they lived in foster care.

Childhood adversity.

All items were either dichotomous by nature, or transformed to be dichotomous where noted; frequencies for the sample are given for each item here as well as in Table 2. Family dysfunction included incarceration of a mother (26.6%), father (35.9%), or sibling (younger and older sibling combined, 17.6%); “parental problem history” with alcohol (21.0%), drugs (18.4%), mental health (9.4%), or physical health (12.0%);, or out-of-home placement (16.4%). Child maltreatment was assessed by history of sexual (13.9%) or physical (23.9%) abuse (both sexual and physical abuse collapsed incidences occurring inside the family and outside of the family); neglect (16.9%); and family conflict, which was based on the respondents answering yes to one of the following (mutually exclusive) experiences: verbal intimidation, “heated arguments,” or exposure to domestic violence (64.9%). Socioeconomic disadvantage included low family income (below $15,000 or below the poverty line, 20.9%), lack of health insurance (5.8%), and a history of parental employment problems (17.2%).

Table 2:

Class and sample proportions endorsing each indicator

Low All Parental Incarceration & Substance Use Poverty & Parental Health Problems High Conflict & High SES High Maltreatment High All Full Sample
Latent class sizes 40.31% 12.04% 13.24% 15.29% 10.97% 8.14%
Maternal incarceration a,b,c,d,e,f,g,h,i,j,k,l,m,n,o 0.074 0.468 0.263 0.126 0.626 0.694 0.266
Paternal incarceration a,b,c,d,e,f,g,h,i,j,k,l,m,n,o 0.175 0.573 0.444 0.348 0.480 0.677 0.359
Sibling incarceration a,b,c,d,e,f,g,h,i,j,k,l,m,n,o 0.140 0.214 0.235 0.144 0.133 0.326 0.176
Parent alcohol abuse a,b,c,d,e,f,g,h,i,j,k,l,m,n,o 0.040 0.761 0.084 0.193 0.043 0.698 0.210
Parental drug use a,b,e,f,i,l,o 0.016 0.716 0.082 0.057 0.028 0.842 0.184
Parental MH problems a,b,c,e,f,g,h,i,j,k,l,m,n,o 0.016 0.111 0.222 0.101 0.028 0.324 0.094
Parent PH problems a,b,c,d,e,f,g,h,i,j,k,l,m,n,o 0.036 0.120 0.390 0.083 0.060 0.249 0.120
Out of home placements b,c,d,e,j,k,l,m,n,o 0.031 0.028 0.077 0.071 0.759 0.535 0.164
Sexual abuse a,b,c,d,e,f,g,h,i,j,k,l,m,n,o 0.047 0.078 0.155 0.224 0.318 0.254 0.139
Physical abuse f,g,h,i,j,k,l,m,n,o 0.033 0.131 0.231 0.519 0.533 0.507 0.240
Neglect j,k,l,m,n,o 0.007 0.030 0.092 0.062 0.766 0.692 0.169
Family conflict a,b,d,e,f,h,i,k,l,o 0.497 0.698 0.768 0.908 0.498 0.833 0.649
Low family income a,b,d,e,f,h,i,k,l,o 0.105 0.232 0.560 0.050 0.154 0.480 0.209
No health insurance a,b,c,d,e,f,g,h,i,j,k,l,m,n,o 0.052 0.062 0.090 0.049 0.040 0.068 0.058
Parental unemployment a,b,e,f,i,k,l,o 0.031 0.267 0.576 0.000 0.026 0.594 0.172
Open in a new tab

Analysis

Latent Class Analysis (LCA; Clogg, 1995) was used to estimate a model that examined diverse patterns of adverse experiences among the sample. Analysis was conducted using Mplus 6.1 (Muthén & Muthén, 2010). All fifteen variables assessing childhood adversity were included as indicators of a latent categorical variable. We estimated the models by incrementally increasing the number of latent classes and comparing indices of fit. Because there is no single fit statistic that may be used to determine the best fitting number of classes, we examined multiple: the log likelihood value, Bayesian Information Criterion (BIC), Lo-Mendell-Rubin test (LMR), and Vuong-Lo-Mendell-Rubin likelihood ratio test (VLMR-LRT). Although the log likelihood values always increase with increasing number of latent classes, BIC statistic takes into consideration the complexity of the model. A lower BIC statistic indicates the better model fit. Both the LMR and the VLMR-LRT compare the fit of a model to the fit of a model with one fewer class (e.g., 4-class model to 3-class model). A significant p value indicates that a model with one more class is a better fitting model. In order to ensure that fitted models were not local solutions, we used random starting values (10 initial-stage iterations; 1000 initial-stage random values; 100 final-stage optimization). The best-fitting model was selected based on these model fit statistics as well as substantive interpretation. After selecting the best fitting model, we re-estimated the model also with 50% random sub-sample. Both the best fitting model and the interpretation of classes remained consistent.

Mplus provides a mechanism to test for class differences on additional variables via the Auxiliary command (Muthén, 2007). This mechanism uses the Wald’s test for mean differences based on class membership in the latent classes, as opposed to assigning cases to classes and testing via ANOVA or similar, which introduces substantial error (Nagin, 1999). We employed this technique to test for differences on demographic variables including living situation.

Results

Table 1 provides the fit statistics for the best-fitting models with one through seven classes. The LMR and VLMR-LRT both suggested that the six-class solution was optimal, however the BIC continued to improve with the addition of a seventh class (the eight-class solution was unreliable due to local maxima). Because of this ambiguity, we examined both the six- and seven-class solutions for interpretability and coherence. The seven-class solution added one small class (5.7%), which did not improve the theoretical meaningfulness of the classes. Thus we retained the six-class solution. The average latent class probabilities for class membership, which are indicators of correct model assignment to the six classes, were good, ranging from .70 to .88. Entropy, which reflects these calculations, was acceptable at 0.69.

Table 1:

Model fit statistics for the 1- through 7-class solutions

Number of Classes Log Likelihood BIC LMR VLMR-LRT
1 −30153.24 151541.33 n/a n/a
2 −28581.38 143836.30 p<0.0001 p<0.0001
3 −27990.35 141096.27 p<0.0001 p<0.0001
4 −27743.45 140114.39 p<0.0001 p<0.0001
5 −27546.66 139309.29 p=0.0059 p=0.0061
6 −27434.75 138932.62 p=0.0001 p=0.0001
7 −27380.52 138818.26 p=0.1212 p=0.1227
Open in a new tab

BIC = Bayesian Information Criterion (BIC), LMR = Lo-Mendell-Rubin test, VLMR-LRT = Vuong-Lo-Mendell-Rubin likelihood ratio test.

The proportions of each latent class that endorsed each indicator are shown in Table 2. A visual depiction of the resultsmay be found in Figure 1. In this figure, we transformed each group’s proportions into z-scores compared to the sample proportions and standard deviations, in order to make a figure that was easily viewable with all indicators at similar magnitudes. The first class had relatively low levels of all ACEs and consisted of the largest portion of the sample (40.3%). Thus we termed this class the Low All class. The second class, with 12.0% of the sample, reported high levels of parental incarceration and substance use, elevated indicators of social disadvantage, but relatively low levels of maltreatment; this class we termed the Parental Incarceration and Substance Use class. The third class contained 13.2% of the sample and was marked by parental health and mental health problems along with very high indicators of social disadvantage; thus this class was termed the Poverty and Parental Health Problems class. The fourth class, in contrast, had relatively good indicators of familial economics, but elevated levels of family conflict and physical abuse. With 15.3% of the sample, we termed this class the High Conflict and High SES class. The fifth class reported very high levels of maltreatment, parental incarceration, and out-of-home placements; however other ACEs were reported less frequently than the sample averages. This class contained 11.0% of the sample and was termed the High Maltreatment class. Finally, the sixth class, which was the smallest at 8.1% of the sample, reported high levels of all indicators of ACEs included in this analysis. We called this class the High All class.

Figure 1.

Figure 1.

Open in a new tab

Latent class profiles compared to sample averages for each indicator. Note: The y-axis represents z-scores compared to sample proportions.

Demographic differences

These classes were also examined for differences on demographic variables (see Table 3). Age differed significantly across classes, although the differences were not large. The Low All class was the oldest with an average of 15.6 years, and the High Maltreatment and High All classes were the youngest, at 15.3 and 15.1 years, respectively. The percentage of females increased from the first through sixth class, with the highest proportion (32.3%) seen in the High All class. Caucasian youth were somewhat concentrated in the Parental Incarceration and Substance Use (66.0%), High Conflict and SES (65.0%), and High All (67.8%) classes. African American youth were more likely to be in the Poverty and Parental Health Problems (31.9%) and High Maltreatment (29.4%) classes. Latino youth were found least frequently in the High Maltreatment (4.4%) and High All classes (4.2%). The Other race class was not significantly different across classes.

Table 3:

Demographic factors across the classes

Low All Parental Incarceration & Substance Use Poverty & Parental Health Problems High Conflict & High SES High Maltreatment High All
Demographic Indicators: Wald test
Age (mean)a,b,d,e,i,j,m,n 15.64 15.44 15.33 15.52 15.25 15.14 45.58***
Femalea,b,c,d,e,g,h,i,l 0.168 0.236 0.257 0.301 0.306 0.323 102.86***
Caucasiana,c,e,f,h,j,l,m,o 0.571 0.660 0.531 0.650 0.580 0.678 47.56***
African Americanb,e,f,h,j,l,m,o 0.263 0.223 0.319 0.226 0.294 0.201 30.18***
Latinod,e 0.073 0.052 0.066 0.052 0.044 0.042 11.95*
Other race 0.125 0.096 0.101 0.102 0.089 0.085 10.83
Living situation:
Both biological parentsa,b,c,d,e,h,i,k,m,n,o 0.220 0.168 0.140 0.159 0.023 0.097 178.22***
Biological mothera,c,f,g,h,i,j,k,n 0.506 0.586 0.628 0.551 0.163 0.582 443.457***
Biological fathera,g,h,i,l 0.124 0.160 0.109 0.132 0.085 0.151 15.206**
Neither biological parenta,d,g,h,i,k,m,o 0.149 0.087 0.123 0.158 0.728 0.170 792.41***
Foster carea,d,e,g,h,i,j,l,m,n,o 0.008 0.001 0.006 0.013 0.284 0.043 222.00***
Open in a new tab
a

Superscripts show significant differences between classes: Low All and Parental Incarceration & Substance Use.

b

Low All and Poverty & Parental Health Problems.

c

Low All and High Conflict & High SES.

d

Low All and High Maltreatment.

e

Low All and High All.

f

Parental Incarceration & Substance Use and Poverty & Parental Health Problems.

g

Parental Incarceration & Substance Use and High Conflict & High SES.

h

Parental Incarceration & Substance Use and High Maltreatment.

i

Parental Incarceration & Substance Use and High All.

j

Poverty & Parental Health Problems and High Conflict & High SES.

k

Poverty & Parental Health Problems and High Maltreatment.

l

Poverty & Parental Health Problems and High All.

m

High Conflict & High SES and High Maltreatment.

n

High Conflict & High SES and High All.

o

High Maltreatment and High All.

Living situation variables differed significantly across the classes (Table 3). A relatively small proportion of any class lived with both biological parents—the highest proportions were seen with the Low All (22.0%) class. The Parental Incarceration and Substance Use, Poverty and Parental Health Problems, and High Conflict and High SES classes had similar patterns with respect to living with biological parents; the majority of these classes lived with their biological mothers, with smaller proportions living with both biological parents or their biological fathers only. The High Placement and Maltreatment class was substantially more likely to be living with neither biological parent (72.8%), and to be living in foster care (28.4%). The High All class was least likely to be living with one or both biological parent(s) of any class except for the High Maltreatment and Placements class.

Discussion

This analysis is among the first to examine heterogeneous patterns of adverse childhood experiences among court-involved youth. Consistent with other studies (Baglivio et al., 2014; Dierkshising et al., 2013), substantial portions of youth reported significant histories of adversity, including multiple forms of childhood maltreatment, parental dysfunction, and socioeconomic disadvantage. Adding to the currently rich literature on cumulative ACEs, our analyses show substantial heterogeneity around childhood adversity, with clusters that appear to have significantly different etiological histories and treatment needs. These clusters add substantially to prior variable-centered analyses that demonstrate aggregate linear trends among adversities and between adversities and outcomes, as well as providing evidence related to adversity composition that add nuance to prior findings regarding level differences (e.g., low, medium, high). We discuss each class in turn, with particular attention to practice implications.

Low Risk Class

This class reported relatively low levels of all ACEs compared to the rest of the sample. In general, they had the lowest reported rate of each adversity, although some ACEs were statistically indistinguishable from other classes. This class also contained fewer females, and a disproportionate number of Latinos and African Americans, which may reflect disparities in policing and punishing certain racial groups in general society. This study sample includes court-involved youth who were identified as moderate-to-high-risk of recidivism during pre-screen assessment based on social and criminal history. Thus, they are certainly at elevated risk compared with normative samples of youth. For example, 17.5% of these youth have a history of paternal incarceration, and 49.7% reported elevated family conflict, underscoring the high adverse experiences exposure of the sample as a whole. Furthermore, although these youth were also most likely to be living with both biological parents, only about one fifth reported living with both biological parents, indicating that few families in any class had a “traditional” structure. Nonetheless, probation officers and clinicians may be able to engage this class’s relatively strong family supports in order to meet these youths’ needs.

Parental Substance Abuse and Incarceration Class

These youth reported high levels of parental incarceration and substance use. They had relatively low levels of maltreatment, and moderate endorsement of social disadvantage. They were more likely to be Caucasian and had the second lowest rate of female membership. Parental crime and incarceration have emerged as experiences that may be especially harmful to the social development of youth. Recent studies have demonstrated that youth with parental incarceration are more likely to experience poverty, perhaps in a cyclical fashion as parents move in and out of the justice system without being able to work (Kjellstrand & Eddy, 2011). Children with an incarcerated parent are more likely to have insecure attachment, especially if their mother is incarcerated, in addition to being at risk of additional negative life experiences, all of which may predispose them to delinquency and other negative outcomes (Murray & Murray, 2010). This class may benefit most from a high-quality mentoring program that would provide both positive socialization and a stable adult influence (Jarjoura, DuBois, Shlafer, & Haight, 2013). Other possible interventions should strengthen family functioning overall, such as the Parent Management Training – Oregon Model, which may be especially effective for families dealing with incarceration (Eddy & Reid, 2002).

Parental Poverty and Health Problems Class

These youth reported high levels of parental health problems and poverty indicators, and they had the highest proportion of African Americans of any of the classes. This portrait of poverty and diminished familial resources is suggestive of a poverty-to-delinquency link, possibly because parents were dealing with significant challenges of their own. This is suggestive of a need for wraparound services that could increase protective resources in terms of economic support, health care, and access to prosocial activities. Although we do not have data about the neighborhoods in which these youth live, the current family economic indicators strongly suggest possible exposure to negative community contexts (e.g., disenfranchised neighborhoods, schools). This class represents overall trends that disproportionately funnel poor youth into the juvenile justice system, via points of contact such as frequent policing in poor neighborhoods and schools (Birckhead, 2012). This is suggestive of a need for wraparound services that could increase protective resources in terms of economic support, health care, and access to prosocial activities (Bruns et al., 2010).

High Conflict/High SES Class

These youth reported the highest levels of exposure to family conflict of any class. They also reported elevated levels of physical abuse. Indicators of social disadvantage suggested relatively better economic situations for these families compared to others in the sample, confirmed by separate analyses that examined for higher income brackets (not shown). We speculate that this class represents profiles of domestic violence in the home, which was supported by analyses (not shown here) regarding physical violence between parents. Links between exposure to domestic violence and externalizing behaviors are well established, especially for boys (Evans, Dallies, & DeLillo, 2008). It can also be related to a number of mental health difficulties, such as anxiety and depression (Berzenski & Yates, 2011). From an interventive standpoint, these youth might have their needs best met via a family practice model, such as functional family therapy (Darnell & Schueler, 2015) that would address family contributors to youth problem behaviors.

High Maltreatment and Placement Class

These youth reported extremely high histories of out-of-home-placements, parental incarceration, physical and sexual abuse, and neglect. These youth were also more likely to be younger in age, African American, and female. They were the least likely to be living with a parent, and by far the most likely to be living in foster care. These youth likely carry substantial effects of traumatic experiences, both in terms of abuse and their histories of familial instability leading to earlier contact with the juvenile justice system. Punitive choices, such as stringent detention, are unlikely to be helpful. Consideration of these youths’ living situation—only a quarter live with either parent—is very important. Whereas many foster care situations are youth-supportive, this is not consistent, and care must be taken to ensure that these youths’ living environments are stable. These dual-system-involved youth may benefit from therapeutic foster care models, such as Treatment Foster Care Oregon (Chamberlain & Reid, 1998). In addition, traumatic stress theories would suggest that these youth need interventions to assist in building coping and social skills, that would counteract the tendency towards hyper-reactivity and hostility in the face of conflict Leve, Chamberlain, & Reid, 2005).

High All Class

These youth reported profound histories of adversity—they reported the highest levels of most indicators of adversity, including family incarceration, parental mental health problems, and drug use. On all indicators for which this class did not have the highest scores of the sample, they had the second highest scores – maltreatment and out-of-home placements were second only to the High Maltreatment and Placements class, and poverty indicators were second only to the Poverty and Parental Health Problems class. This global picture of ACEs is suggestive of a serious need for multifaceted interventions to address trauma, poor resources, and global difficulties in functioning. It is also important to note that this group was the youngest and most likely female. Given early onset leading to further involvement in the juvenile justice system, effective interventions targeting this group can have a greater impact in the long run. Furthermore, despite the heavier involvement of boys in the justice system, our findings suggest that girls in the juvenile justice system exhibit the highest risk. As with other classes, trauma-informed interventions are likely necessary, especially inclusive of girls, to interrupt negative trajectories with these youth. Although these youth predominantly live with a family member, the data suggest that this family environment is not likely to be a positive one. In order to achieve success with these youth, multifaceted interventions that incorporate a family component would be important. Studies have shown that youth with the most need benefit the most from interventions that address multiple domains simultaneously (Farmer, Farmer, Estell, & Hutchings, 2007), such as Multisystemic Therapy (Henggeler, Mihalic, Rone, Thomas, & Timmons-Mitchell, 1998). Clinicians would also need to be well trained to be empathetic with youth who show the most severe behavioral problems.

Limitations

This study has limitations worth noting. The data derive from an assessment completed by probation officers, to whom court-involved youth may be less likely to report experiences of maltreatment and adversity. Nevertheless, Washington State has made adequate implementation of this assessment a high priority, including extensive training of probation officers using the tool (Barnoski, 2004b). Studies have also shown that a related risk tool, the Florida Positive Achievement Change Tool, had strong interrater reliability with different types of staff delivering the assessment (Winokur-Early, Hand, & Blankenship, 2012)

Although the retrospective nature of this data necessitates caution, multiple studies have demonstrated that retrospective reports correlate strongly with other sources of verified data (Smith, Ireland, Thornberry, & Elwyn, 2008). As respondents here are adolescents, the time period of retrospection is much shorter than for adult samples that have found adequate variance and stable linear trends of ACEs with health and functioning outcomes even with lengthy retrospection periods (Hardt & Rutter, 2004; Yancura & Aldwin, 2009). This assessment is more epidemiologic than clinical in nature, aiming to identify cumulative exposure across an established set of adversities. The chronicity or severity of these exposures is, thus, not captured which limits ability to assess differences such as magnitude of maltreatment association with.future outcomes (Smith & Thornberry, 1995), However, established short-form adversity assessments are feasible for routine pediatric screening, can provide important information for service providers, and provide opportunities for merging administrative data across systems to gain a fuller picture of the etiology and trajectories of early life adversity (Murphy et al., 2014; Putnam-Hornstein, Needell, & Rhodes, 2013 ).

Additionally, these data are taken from one county in Washington State, which may constrain generalizability. However, the present sample is reasonably diverse compared with many US jurisdictions. This particular county contains urban, suburban, and rural regions, as well as Native American reservations, within its boundaries. Finally, this study lacks assessment of the neighborhood and community contexts in which youth reside. Our indicators of family social disadvantage are likely to be correlated with neighborhood poverty. However, we are not able to test for independent effects net of individual or family contexts relative to community-level characteristics. Future research should expand on the multilevel effects of adversity among court-involved youth.

Conclusions

These analyses add to a growing body of literature that suggests that court-involved youth carry substantial histories of adversity and that greater cumulative adversity is associated with more negative health and development, extending prior findings and theorizing about ACEs (Wolff et al., 2015) These backgrounds of trauma, social disadvantage, and other adversities carry information about risk and protective factors that are imperative to consider when selecting and implementing interventions to prevent recidivism and to improve outcomes. Moreover, these analyses provide strong support for the use of risk assessment tools to target interventions for differing needs of court involved youth. Simply using the risk assessment tool to determine whether or not the youth is “high” risk for reoffending might lead to more punitive approaches. The results of this study illustrate how risk assessment tools can be used to uncover the particular needs of youth, particularly around histories of trauma and adversity. Although the recent push to transform systems of care to be trauma-informed (Ko et al., 2008) is noteworthy, more specified information could provide clinicians guidance as to which types of interventions will benefit youth most.

Furthermore, even in jurisdictions that have focused on providing trauma-informed care and implementing evidence-based programs, many regions do not make evidence-based programs available for court-involved youth unless they are in detention settings such as residential treatment programs (e.g, Ford & Blaustein, 2013). This is unfortunate, particularly given that court-involvement (e.g., probation) on its own is likely to increase the rates of recidivism and involvement with the adult criminal justice system (Gatti, Trembly, & Vitaro, 2009). The findings of this study underscore the need to address histories of trauma and adversity among court-involved youth across community- and detention-settings.

Acknowledgments

This research was supported in part by a grant from the National Institute on Mental Health grant 5 T32 MH20010 “Mental Health Prevention Research Training Program”, the National Center For Advancing Translational Sciences of the National Institutes of Health under Award Number TL1TR000422, and a Eunice Kennedy Shriver National Institute of Child Health and Human Development research infrastructure grant, R24 HD042828, to the Center for Studies in Demography & Ecology at the University of Washington.

Contributor Information

Patricia Logan-Greene, University at Buffalo.

B.K. Elizabeth Kim, University of Southern California.

Paula S. Nurius, University of Washington

References

  1. Abram KM, Teplin LA, Charles DR, Longworth SL, McClelland GM, & Dulcan MK (2004). Posttraumatic stress disorder and trauma in youth in juvenile detention. Archives of general psychiatry, 61(4), 403–410. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Anda RF, Butchart A, Felitti VJ, & Brown DW (2010). Building a framework for global surveillance of the public health implications of adverse childhood experiences. American Journal of Preventive Medicine, 39(1), 93–98. [DOI] [PubMed] [Google Scholar]
  3. Anda RF, Felitti VJ, Bremner JD, Walker J, Whitfield C, & Perry B (2006). The enduring effects of abuse and related adverse experiences in childhood: A convergence of evidence from neurobiology and epidemiology. European Archives of Psychiatry & Clinical Neurosciences, 256(3), 174–186. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Baglivio MT, & Epps N (2015). The interrelatedness of adverse childhood experiences among high-risk juvenile offenders. Youth Violence & Juvenile Justice, 1–20. doi: 10.1177/1541204014566286. [DOI] [Google Scholar]
  5. Baglivio MT, Epps N, Swartz K, Sayedul Huq M, Sheer A, & Hardt NS (2014). The prevalence of adverse childhood experiences (ACE) in the lives of juvenile offenders. Journal of Juvenile Justice, 3(2), 1–23. [Google Scholar]
  6. Baglivio MT, & Jackowski K (2013). Examining the validity of a juvenile offending risk assessment instrument across gender and race/ethnicity. Youth Violence and Juvenile Justice, 11(1), 26–43. [Google Scholar]
  7. Baglivio MT, Wolff KT, Epps N, & Nelson R (2015). Predicting adverse childhood experiences: The importance of neighborhood context in youth trauma among delinquent youth. Crime & Delinquency. 0011128715570628. [Google Scholar]
  8. Baglivio MT, Wolff KT, Piquero AR, & Epps N (2015). The relationship between adverse childhood experiences (ACE) and juvenile offending trajectories in a juvenile offender sample. Journal of Criminal Justice, 43(3), 229–241. [Google Scholar]
  9. Barnoski R (2004a). Assessing risk for re-offense: Validating the Washington State Juvenile Court assessment appendices. Olympia, WA: Washington State Institute for Public Policy. [Google Scholar]
  10. Barnoski R (2004b) Washington State Juvenile Court Assessment Manual, Version 2.1. Olympia: Washington State Institute for Public Policy. [Google Scholar]
  11. Bergman LR, & Trost K (2006). The person-oriented versus the variable-oriented approach: Are they complementary, opposites, or exploring different worlds?. Merrill-Palmer Quarterly, 52(3), 601–632. [Google Scholar]
  12. Berzenski SR, & Yates TM (2011). Classes and consequences of multiple maltreatment: A person-centered analysis. Child Maltreatment, 16(4), 250–261. [DOI] [PubMed] [Google Scholar]
  13. Birckhead TR (2012). Delinquent by reason of poverty. Washington University Journal of Law & Policy, 38, 53–107. [Google Scholar]
  14. Bruns EJ, Walker JS, Zabel M, Matarese M, Estep K, Harburger D, ... & Pires SA, (2010). Intervening in the lives of youth with complex behavioral health challenges and their families: The role of the wraparound process. American Journal of Community Psychology, 46(3–4), 314–331. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Chamberlain P, & Reid J (1998). Comparison of two community alternatives to incarceration for chronic juvenile offenders. Journal of Consulting and Clinical Psychology, 5, 857–863. [DOI] [PubMed] [Google Scholar]
  16. Clogg CC (1995). Latent class models In Handbook of statistical modeling for the social and behavioral sciences (pp. 311–359). Springer US. [Google Scholar]
  17. Cronholm PF, Forke CM, Wade R, Bair-Merritt MH, Davis M, Harkins-Schwarz M, ... & Fein JA (2015). Adverse childhood experiences: Expanding the concept of adversity. American Journal of Preventive Medicine, 49(3), 354–361. [DOI] [PubMed] [Google Scholar]
  18. Darnell AJ, & Schuler MS (2015). Quasi-experimental study of Functional Family Therapy effectiveness for juvenile justice aftercare in a racially and ethnically diverse community sample. Children & Youth Services Review, 50, 75–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. DeNavas-Walt C (2010). Income, poverty, and health insurance coverage in the United States (2005). DIANE Publishing. [Google Scholar]
  20. Dierkhising CB, Ko S, Goldman HJ (2013). Trauma-informed juvenile justice roundtable: Current issues and directions in creating trauma-informed juvenile justice systems. Los Angeles, CA & Durham, NC: National Center for Child Traumatic Stress. [Google Scholar]
  21. Dong M, Anda RF, Felitti VJ, Dube SR, Williamson DF, Thompson TJ, Loo CM, & Giles WH (2004). The interrelatedness of multiple forms of childhood abuse, neglect, and household dysfunction. Child Abuse and Neglect, 28, 771–784. [DOI] [PubMed] [Google Scholar]
  22. Duke NN, Pettingell SL, McMorris BJ, & Borowsky IW (2010). Adolescent violence perpetration: Associations with multiple types of adverse childhood experiences. Pediatrics, 125(4), e778–e786. [DOI] [PubMed] [Google Scholar]
  23. Eddy JM, & Reid JB (2002). The antisocial behavior of the adolescent children of incarcerated parents: A developmental perspective. Washington, DC: American Psychological Association; Retrieved from https://aspe-dev.akira-tech.com/sites/default/files/pdf/74986/eddy.pdf [Google Scholar]
  24. Evans SE, Davies C, & DiLillo D (2008). Exposure to domestic violence: A meta-analysis of child and adolescent outcomes. Aggression and violent behavior, 13(2), 131–140. [Google Scholar]
  25. Farmer TW, Farmer EMZ, Estell DB, & Hutchings BC (2007). The developmental dynamics of aggression and the prevention of school violence. Journal of Emotional Behavioral Disorders, 15(4), 197–208. [Google Scholar]
  26. Felitti VJ, Anda RF, Nordenberg D, Williamson DF, Spitz AM, Edwards V, ... & Marks JS (1998). Relationship of childhood abuse and household dysfunction to many of the leading causes of death in adults: The Adverse Childhood Experiences (ACE) Study. American Journal of Preventive Medicine, 14(4), 245–258. [DOI] [PubMed] [Google Scholar]
  27. Ford JD, & Blaustein ME (2013). Systemic self-regulation: a framework for trauma-informed services in residential juvenile justice programs. Journal of Family Violence, 28(7), 665–677. [Google Scholar]
  28. Ford J, Chapman J, Hawke J, & Albert D (2007). Trauma among youth in the juvenile justice system: Critical issues and new directions (National Center for Mental Health and Juvenile Justice Research Brief). Retrieved June 6, 2009. [Google Scholar]
  29. Fox BH, Perez N, Cass E, Baglivio MT, & Epps N (2015). Trauma changes everything: Examining the relationship between adverse childhood experiences and serious, violent and chronic juvenile offenders. Child Abuse & Neglect, 46, 163–173. [DOI] [PubMed] [Google Scholar]
  30. Gatti U, Tremblay RE, & Vitaro F (2009). Iatrogenic effect of juvenile justice. The Journal of Child Psychology and Psychiatry, 50(8). 991–998. [DOI] [PubMed] [Google Scholar]
  31. Greenwald R (2014). Trauma and juvenile delinquency: Theory, research, and interventions. Routledge. [Google Scholar]
  32. Grevstad JA (2010, June 8). Adverse childhood experiences and juvenile justice PowerPoint delivered to Washington State Family Policy Council, Walla Walla, WA. [Google Scholar]
  33. Hardt J, & Rutter M (2004). Validity of adult retrospective reports of adverse childhood experiences: review of the evidence. Journal of Child Psychology and Psychiatry, 45(2), 260–273. [DOI] [PubMed] [Google Scholar]
  34. Henggeler SW, Mihalic SF, Rone L, Thomas C, & Timmons-Mitchell J (1998). Blueprints for violence prevention, book six: Multisystemic therapy. Boulder, CO: Center for the Study and Prevention of Violence. [Google Scholar]
  35. Jarjoura GR, DuBois DL, Shlafer RJ, & Haight KA (2013). Mentoring children of incarcerated parents: A synthesis of research and input from the listening session held by the office of juvenile justice and delinquency prevention and the white house domestic policy council and office of public engagement. Office of Juvenile Justice and Delinquency Prevention. [Google Scholar]
  36. Kjellstrand JM, & Eddy JM (2011). Parental incarceration during childhood, family context, and youth problem behavior across adolescence. Journal Of Offender Rehabilitation, 50(1), 18–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Ko SJ, Ford JD, Kassam-Adams N, Berkowitz SJ, Wilson C, Wong M, ... & Layne CM (2008). Creating trauma-informed systems: child welfare, education, first responders, health care, juvenile justice. Professional Psychology: Research and Practice, 39(4), 396. [Google Scholar]
  38. Lanza ST, Rhoades BL. Latent Class Analysis: An alternative perspective on subgroup analysis in prevention and treatment. Prevention Science, 14(2), 157–168. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Larkin H, Shields JJ, & Anda RF (2012). The health and social consequences of Adverse Childhood Experiences (ACE) across the lifespan: An introduction to prevention and intervention in the community. Journal of Prevention & Intervention in the Community, 40(4), 263–270. [DOI] [PubMed] [Google Scholar]
  40. Leve LD, Chamberlain P, & Reid JB (2005). Intervention outcomes for girls referred from juvenile justice: effects on delinquency. Journal of Consulting & Clinical Psychology, 73(6), 1181. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Logan-Greene P, Green S, Nurius P, & Longhi D (2014). Distinct contributions of adverse childhood experiences and resilience resources: A cohort analysis of adult physical and mental health. Social Work in Health Care, 53(8), 776–797. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. McCutcheon AL (1987). Latent class analysis (No. 64). Sage. [Google Scholar]
  43. Mersky JP, Topitzes J, & Reynolds AJ (2013). Impacts of adverse childhood experiences on health, mental health, and substance use in early adulthood: A cohort study of an urban, minority sample in the US. Child Abuse & Neglect, 37(11), 917–925. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Mulder E, Vermunt J, Brand E, Bullens R, & Marle H (2012). Recidivism in subgroups of serious juvenile offenders: different profiles, different risks?. Criminal Behaviour and Mental Health, 22(2), 122–135. [DOI] [PubMed] [Google Scholar]
  45. Murphy A, Steele M, Dube SR, Bate J, Bonuck K, Meissner P, ... & Steele H (2014). Adverse childhood experiences (ACEs) questionnaire and adult attachment interview (AAI): Implications for parent child relationships. Child Abuse & Neglect, 38(2), 224–233. [DOI] [PubMed] [Google Scholar]
  46. Murray J, & Murray L (2010). Parental incarceration, attachment and child psychopathology. Attachment & human development, 12(4), 289–309. [DOI] [PubMed] [Google Scholar]
  47. Muthén LK (2007). Mplus: Statistical Analysis with Latent Variables (Version 4.21) [Computer software]. [Google Scholar]
  48. Muthén LK, & Muthén BO (2010). Mplus User’s Guide: Statistical Analysis with Latent Variables: User’s Guide. Muthén & Muthén. [Google Scholar]
  49. Nagin DS (1999). Analyzing developmental trajectories: A semiparametric, group-based approach. Psychological Methods, 4(2), 139–157. [DOI] [PubMed] [Google Scholar]
  50. Nurius PS, Green S, Logan-Greene P, & Borja S (2015). Life course pathways of adverse childhood experiences toward adult psychological well-being: A stress process analysis. Child Abuse & Neglect, 45, 143–153. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Putnam FW (2006). The impact of trauma on child development. Juvenile and Family Court Journal, 57(1), 1–11. [Google Scholar]
  52. Putnam-Hornstein E, Needell B, & Rhodes AE (2013). Understanding risk and protective factors for child maltreatment: The value of integrated, population-based data. Child Abuse & Neglect, 37(2), 116–119. [DOI] [PubMed] [Google Scholar]
  53. Sacks V, Murphey D, & Moore K (2014). Adverse childhood experiences: National and state level prevalence. Research Brief Child Trends, Publication #2014–28. http://www.childtrends.org/wp-content/uploads/2014/07/Brief-adverse-childhood-experiences_FINAL.pdf [Google Scholar]
  54. Shonkoff JP, Garner AS, Siegel BS, Dobbins MI, Earls MF, McGuinn L, ... & Wood DL (2012). The lifelong effects of early childhood adversity and toxic stress. Pediatrics, 129(1), e232–e246. [DOI] [PubMed] [Google Scholar]
  55. Skowyra KR, & Cocozza JJ (2007). Blueprint for change: A comprehensive model for the identification and treatment of youth with mental health needs in contact with the juvenile justice system. Policy Research Associates, Inc. [Google Scholar]
  56. Smith CA, Ireland TO, Thornberry TP, & Elwyn L (2008). Childhood maltreatment and antisocial behavior: comparison of self-reported and substantiated maltreatment. American Journal of Orthopsychiatry, 78(2), 173–186. [DOI] [PubMed] [Google Scholar]
  57. Smith C, & Thornberry TP (1995). The relationship between childhood maltreatment and adolescent involvement in delinquency. Criminology, 33(4), 451–481. [Google Scholar]
  58. Wade R, Shea JA, Rubin D, & Wood J (2014). Adverse childhood experiences of low-income urban youth. Pediatrics, 134(1), e13–e20. [DOI] [PubMed] [Google Scholar]
  59. Washington State Institute for Public Policy (2004). Assessing risk for re-offense: Validating the Washington State Juvenile Court assessment. Olympia, WA. [Google Scholar]
  60. Winokur-Early K, Hand GA, & Blankenship J (2012). Validity and reliability of the Florida Positive Achievement Change Tool (PACT) risk and needs assessment instrument: A three-phase evaluation (Validation study, factor analysis, inter-rater reliability). Tallahassee, FL: Justice Research Center. Retrieved from http://www.djj.state.fl.us/docs/probation-policy-memos/jrc-comprehensive-pact-validity-and-reliability-study-report-2012.pdf?Status=Master&sfvrsn=2 [Google Scholar]
  61. Wolff KT, Baglivio MT, & Piquero AR (2015). The relationship between adverse childhood experiences and recidivism in a sample of juvenile offenders in community-based treatment International Journal of Offender Therapy and Comparative Criminology. Published online before print November 12, 2015, doi: 10.1177/0306624X15613992 [DOI] [PubMed] [Google Scholar]
  62. Yancura LA, & Aldwin CM (2009). Stability and change in retrospective reports of childhood experiences over a 5-year period: findings from the Davis Longitudinal Study. Psychology & Aging, 24(3), 715 721. [DOI] [PubMed] [Google Scholar]

RESOURCES