Maternal Criminal Justice and Child Welfare Involvement: Associations between Risk Exposures and Childhood Mental Health

Over the past 30 years, arrest and incarceration rates for women have steadily increased; as a result, children of mothers involved in the criminal justice system constitute a rapidly growing at-risk population (Glaze & Maruschak, 2008; Maruschak, Glaze, & Mumola, 2010). Children with such mothers encounter numerous challenges, one of which is vulnerability to the child welfare system (Hairston, 2008; Johnson & Waldfogel, 2002). Dual involvement with child welfare and criminal justice systems presents a unique phenomenon: children are potentially exposed to multiple vulnerabilities that can contribute to maladaptive functioning. Although studies have provided important data on children of incarcerated parents, the research evidence is unclear on how exposure to maternal criminal justice involvement (MCJI) and co-occurrences of other maternal and environmental risks influence the mental health of children who are assessed for maltreatment.

This study explored the associations between MCJI and maternal risks (substance abuse, mental illness) and environmental risks (adverse community conditions, violence exposure) among children assessed for maltreatment. Moreover, the study examined whether exposure to MCJI, maternal risks, and environmental risks affect children's risk of adverse mental health outcomes (for example, internalizing or externalizing problems). Finally, the study investigated whether MCJI moderates the association between maternal and environmental risks and children's mental health outcomes. Examining how MCJI and maternal and environmental risks influence children's mental health is essential to understanding the implications for child welfare and criminal justice practices, policies, and future research.

MCJI

Regardless of whether a mother is arrested without further adjudication, sanctioned to probation, or sentenced to a term of imprisonment, MCJI is a nonnormative event with the potential to negatively affect children's well-being. A review of the literature reveals a range of mental health concerns associated with MCJI. In terms of development and mental health, specifically, children's exposure to MCJI has been associated with internalizing and externalizing outcomes, such as developmental regression, inappropriate coping strategies, somatization, depression, aggression, shame, guilt, and attachment-related issues (Baker, McHale, Strozier, & Cecil, 2010; Johnson & Waldfogel, 2002; Kampfner, 1995; Myers, Smarsh, Amlund-Hagen, & Kennon, 1999; Poehlmann, 2005). Kampfner's (1995) study reported that approximately 75% of the children in the sample exhibited maladaptive symptoms, such as depression, problems concentrating, difficulty sleeping, and vivid memories associated with their mother's involvement with law enforcement and the judicial system. In another study, Poehlmann (2005) found that 40% of children with incarcerated mothers exhibited extreme emotional and behavioral reactions, which were attributed to the experience of maternal incarceration. Many children respond in troubling ways and experience a range of negative outcomes as a result of their mother's involvement with the criminal justice system. These experiences are compounded in many cases by children's exposure to additional stressors or risk factors beyond simply MCJI; many of these children also encounter adverse maternal and environmental risks that have the potential to increase children's vulnerabilities to poor outcomes.

MCJI and Co-Occurring Maternal and Environmental Risks

Women who come to the attention of the criminal justice system often experience barriers and challenges in socioecological domains outside the criminal context, many of which complicate their involvement with the justice system. For example, arrested and incarcerated individuals are disproportionately affected by substance abuse and mental health issues as compared with the general population (Fazel, Bains, & Doll, 2006; Fazel & Danesh, 2002; Hammett, Roberts, & Kennedy, 2001). Glaze and Maruschak (2008) surveyed prison inmates in 2004, finding that 72.8% of incarcerated mothers experienced mental health issues and 63.6% had a history of substance abuse. Mothers with criminal justice involvement are also disproportionately affected by exposure to violence and noxious environmental conditions (DeHart, 2008; Raj et al., 2008).

Researchers studying how these and other parental and environmental factors affect childhood outcomes have demonstrated that, like MCJI, these conditions adversely affect children. For example, children exposed to parental substance abuse, parental mental illness, violence, and community adversity are at higher risk of adverse emotional and behavioral development, including children assessed for maltreatment (Black & Krishnakumar, 1998; Conners et al., 2003; Dore, Doris, & Wright, 1995; Hammen & Brennan, 2003; Lindsey et al., 2008; McCrae, Chapman, & Christ, 2006). Studies have also found that exposure to parental substance abuse and mental illness is prevalent among children with criminal justice involved parents (Phillips, Burns, Wagner, & Barth, 2004; Phillips, Erkanli, Keeler, Costello, & Angold, 2006).

Examining environmental factors, Mackintosh, Myers, and Kennon (2006) found that violence and neighborhood dangers were common among children of incarcerated mothers. Of the children in the study, 36% reported witnessing violence, 27% reported that their neighborhoods posed too many risks for them to play safely, and 25% had to hide to protect themselves from neighborhood dangers (shootings). Studying children exposed to violence, DeHart and Altshuler (2009) found that children of incarcerated mothers exhibited elevated internalizing and externalizing behaviors as compared with their peers with different family histories. The children who witnessed violence were likely to be passive and frequently cry or become verbally and physically aggressive. Children's exposure to violent or unsafe neighborhoods and other negative stimuli introduces complications for mental health, which are compounded by MCJI.

Maternal Criminal Justice and Child Welfare Involvement

In addition to the maternal and environmental risk factors, the circumstances of children with MCJI who are assessed by child welfare agencies merit special consideration. Children assessed for maltreatment are exposed to a variety of factors that influence their mental health—factors beyond the suspected abuse or neglect. Children assessed for maltreatment are at greater risk for mental health problems than are children in the general population (Leslie et al., 2005; Manly, Kim, Rogosch, & Cicchetti, 2001; McCrae et al., 2006; Moylan et al., 2010; Schneiderman & Villagrana, 2010). Research has found that children who experience maltreatment are more likely to exhibit internalizing and externalizing problems that manifest as anger, depression, anxiety, and disassociation (English et al., 2005; Ford, Wasser, & Conner, 2011; Johnson et al., 2002).

Turning directly to the intersection of MCJI and child welfare involvement, according to a nationally representative study, approximately 11.3% of children who came to the attention of child welfare agencies had mothers with an arrest history (Phillips et al., 2004). In another study, Phillips, Leathers, and Erkanli (2009) found that at baseline 8% to 14% of children reported to child welfare agencies came from families with probationed parents. Of these, 96% were children with probationed mothers and had clinically significant emotional and behavioral problems, compared with 10% of children in the general population who exhibit these same behaviors. In addition to the unique challenges many children with MCJI encounter, including co-occurring maternal and environmental risks, the intersection of the child welfare and criminal justice systems theoretically increases their susceptibility to poor mental health outcomes.

The current study builds on previous research and addresses gaps in the literature. Based on the literature, this study hypothesized that children who experience MCJI, maternal substance abuse, maternal mental illness, violence exposure, and adverse community conditions will have increased levels of internalizing and externalizing problems. Children of criminal justice involved mothers were predicted to be more likely to have mothers with substance abuse and mental illness and experience higher levels of violence exposure and adverse community conditions. The study also hypothesized that MCJI moderates the associations between maternal and environmental risks and childhood mental health, such that children who experience the co-occurrence of MCJI with other maternal or environmental risks would be more likely to experience internalizing and externalizing problems than children who experience only maternal or environmental risks.

Method

Sample Design and Procedures

This study used data from the National Survey on Child and Adolescent Well-Being (NSCAW) Child Protective Services (CPS), a national probability longitudinal study of 5,501 youth and families assessed for child maltreatment. The NSCAW research design used a two-staged stratified cluster sample of children ages 0 to 14 in the United States. A number of the children in the NSCAW study turned 15 (n = 99) after data collection began; however, the official ages for recruitment were 0 to 14. In Stage 1, nine sampling strata (eight strata represented by the eight states with the largest child welfare caseloads and a ninth stratum corresponding to the District of Columbia and 36 states) were divided into 92 randomly selected primary sampling units (PSU) using a probability- proportionate-to-size procedure. In the second stage, 5,501 children were selected from the 92 PSU and stratified by age, receipt of child welfare services, type of placement, and type of maltreatment. Cases were excluded in the following circumstances: the child was older than 14 years of age, a sibling was participating in the study, the child was identified as the perpetrator of maltreatment, or the CPS referral was deemed insufficient to investigate. Wave 1 of the NSCAW study began in October 1999; waves 2 to 4 occurred at 12 months, 18 months, and 36 months after the initial investigation; and wave 5 follow-up was completed 59 to 97 months after the investigation. Additional details of the study design are available in Dowd et al. (2006).

The sample for this study included children ages five to 15, residing with their biological mothers in an in-home setting (that is, non-foster care setting) at the time of data collection. Data were taken from wave 1 interviews with biological mothers who had provided information regarding their criminal justice status. Of the 5,501 children in the CPS sample, 61.6% resided with their biological mothers (n = 3,387) and 52.6% (n = 1,780) of those children were between five and 15 years of age. Of the 1,780 children, 97.5% had completed information on maternal criminal justice status. The total sample for the current study is 1,735 children.

Measures

MCJI

MCJI was a researcher-constructed variable gleaned from Audio Computer-Assisted Self Interview (ACASI) technology generated questions. The ACASI was administered to children's primary caregivers (in this sample, primary caregivers were biological mothers) to elicit information about criminal justice status. The researcher-constructed involvement category was a collapsed code defined by whether a mother self-reported an arrest, sanctioned probation, or incarceration history after her child's birth. The mother's criminal justice status was a categorical measure: 0 = non-involvement and 1 = involvement. An involvement code was given only if a mother's involvement with the criminal justice system occurred after her child's birth. If a mother had criminal justice involvement prior to her child's birth with no subsequent contact with the criminal justice system, a non-involvement code was assigned. Cases with missing data on maternal criminal justice status, 2.5% (n = 45), were excluded from the study sample.

Maternal Risks

The Composite International Diagnostic Interview Short Form (CIDI-SF) ACASI-administered questions generated data on maternal substance abuse and mental illness. The CIDI-SF is a highly standardized measure that consists of items derived from the World Health Organization's CIDI, a well-established assessment tool based on the National Comorbidity Survey data (Kessler et al., 1994). The CIDI-SF items used to assess maternal substance abuse included questions pertaining to whether the mother received treatment for alcohol or drug abuse or required ongoing substance abuse services within the past 12 months. Maternal substance abuse was a categorical measure: 0 = no substance abuse and 1 = substance abuse. Because of low “yes” responses for drug abuse in particular, an affirmative response to any of the CIDI-SF alcohol or drug abuse questions was coded as a 1 response. This decision is congruent with the intention to measure substance abuse, rather than use of a composite score that meets diagnostic criteria for clinical alcohol or drug dependence. The CIDI-SF items used to assess maternal mental illness included questions concerning whether the mother required ongoing mental health services, received services from a mental health facility, or were prescribed mental health medications within the past 12 months. Maternal mental illness was also a categorical measure: 0 = no mental illness and 1 = mental illness. The internal consistency for the CIDI alcohol and drug abuse items has ranged from .70 to .94 (Cottler et al., 1991; Ustun et al., 1997; Wittchen, 1994). Test–retest statistics for the alcohol, drug, and mental illness items have had kappas of .62 to .78 (Wittchen, 1994).

Environmental Risks

To measure environmental risk factors, specifically those related to violence and neighborhood safety, this study relied on two scales. The first, the Violence Exposure Scale-Revised (VEX-R), is a 23-item child self-report scale that uses a four-point Likert-type scale to measure children's exposure to violence (as a witness to or recipient) in four domains: home, school, neighborhood, and television watching (Raviv et al., 2001). The scale distinguishes between mild violence (yelled at, beat up, pushed, chased) and severe violence (witnessed an arrest, robbed, threatened with a gun or knife). This study used the mild/severe total exposure to violence scores for analyses. Children ages five to 10 years were administered the VEX-R pictorial cartoon-based interview and asked if they had ever witnessed or experienced events depicted in the cartoons. Children age 11 and older were asked the same questions without the pictorial depiction of events. The response categories (1 = never; 2 = once; 3 = a few times; and 4 = a lot of times) were recoded to reflect the instances of violence exposure: 0 = never; 1 = once; 2 = a few times; and 3 = lots of times, with a theoretical range of 0 to 19 for mild/severe violence exposure. Higher scores indicated increased exposure to violence. The Cronbach's alpha for the VEX-R has ranged from .72 to .96 (Dowd et al., 2006; Raviv, Raviv, Shimoni, Fox, & Leavitt, 1999; Shahinfar, Fox, & Leavitt, 2000).

To establish a measure of environmental and neighborhood safety, this study used items 1 through 5 from the NSCAW research team's nine-item questionnaire, an abridged version of the Community Environment Scale (CES). The CES reports on parents' perceptions of the severity of hazardous conditions present in their communities at the time of the interview. This study used items 1 through 5 of the NSCAW CES. The questions included hazardous indicators such as assaults, muggings, the presence of delinquent gangs, and drug activity within their communities. The NASW CES used a three-point Likert-type scale to measure parents' assessment of these problems in their local contexts: 0 = not a problem at all; 1 = somewhat of a problem; and 2 = a big problem. The theoretical total score ranged from 0 to 10. Higher NSCAW CES total scores indicate a higher perceived level of adverse community conditions. Internal consistency in the NSCAW sample for the CES is .86 (Dowd et al., 2006).

Childhood Mental Health

The Child Behavior Checklist's (CBCL) (Achenbach, 1991) two broadband problem scales, internalizing (anxiety, depression, withdrawal) and externalizing (aggression, impulsivity, delinquency), were used in this study. The CBCL is an empirically supported 113-item measure with well-established reliability and validity, which was administered to children's identified parental caregivers to measure childhood mental health (Achenbach, 1991). Parents responded to items pertaining to the frequency of their child's behaviors, using a three-point Likert-type scale: 1 = not true; 2 = somewhat or sometimes true; and 3 = very true or often true. Responses were recoded as: 0 = not true; 1 = somewhat or sometimes true; and 2 = very true or often true. Higher T scores on the two broadband problem scales indicate increased pathology. T scores between 61 and 63 are in the borderline clinical range and scores greater than 63 are in the clinical range. The Cronbach's alpha for the CBCL has ranged from .78 to .97 (Achenbach, 1991; Achenbach & Rescorla, 2001).

Control Variables

The study used five control variables: sex, age, race, substantiated abuse or neglect, and income. Sex was a dichotomous variable (male = 0 and female = 1). The age variable ranged from five to 15 years. Race was coded as a dichotomous dummy variable, where 0 = all other races (that is, white, Hispanic, other) and 1 = black. Substantiated abuse or neglect was a dichotomous variable (0 = no substantiated abuse or neglect and 1 = substantiated abuse or neglect). Income was a categorical variable, measured in $10,000 increments with a range from $0 to $40,000, and greater.

Data Analysis

Analyses were conducted with weighted data to adjust for the complex sampling design and to estimate population parameters that closely represent youth assessed for maltreatment in the U.S. child welfare system. Details on the construction of NSCAW weights are available in Dowd et al. (2006). Correlation analyses examined the associations between MCJI and maternal and environmental risks. The study also used correlations to analyze the associations between the predictor variables (that is, MCJI, maternal and environmental risks) and outcome variables (that is, internalizing or externalizing behaviors).

The research questions were tested using a series of regression analyses. Regression analyses included all the predictor variables that significantly correlated with the outcome variables. A stepwise approach with a mathematical criterion was used to establish the order in which the predictors were entered into each model. In the first model, the variable with the highest zero-order correlation with the outcome was selected. The second model retained the first predictor, and chose the second predictor by selecting the variable with the largest semi-partial correlation with the outcome. Subsequent models were built by selecting the next highest partial-correlation. A strength of this method is that the stepwise method assesses for redundancy and removes unnecessary predictors for each new variable added to a model, resulting in a final model with the greatest predictive power (Field, 2009) (see Tables 2 and 3).

This study followed Baron and Kenny's (1986) model for establishing moderation. First, to test for moderation, the study analyzed predictor variables (maternal substance abuse, community adversity) to determine whether they were significantly associated with the outcome variables (internalizing or externalizing behaviors). Second, the hypothesized moderating variable (MCJI) was analyzed to test for a significant association with the outcome variables. Finally, to examine whether MCJI moderated the associations between maternal and environmental risks and the two outcomes variables, interaction terms (MCJI × maternal substance abuse, MCJI × community adversity) were created and tested for a significant association with the outcome variables in a final regression analysis (see Table 4).

Results

Descriptive Statistics

The study sample consisted of 1,735 children who were assessed by child protective services and resided with their mothers in in-home settings. Boys made up 51.4% of the sample. Children's mean age was 8.3 (SD = 2.9). The reported race/ethnic composition of the sample was 46.1% white, 27.1% black, 18.8% Hispanic, and 8.0% other. Among the children in the sample, 22.5% had experienced maternal criminal involvement during their lifetimes, 12.1% had a mother with a history of mental illness, and 2.7% had a mother with a history of substance abuse. The CES community adversity had a mean of 1.86 (SD = 2.49). The exposure to violence mean score was 5.37 (SD = 3.85). The CBCL Internalizing mean score was 54.31 (SD = 11.80), with 4.4% of the sample scoring in the borderline clinical range and 19.1% scoring in the clinical range. The CBCL Externalizing mean score was 57.35 (SD = 11.86), with 5.4% of the sample scoring in the borderline clinical range and 29.6% scoring in the clinical range.

Bivariate Correlations

Predictors and Children's Mental Health

The children's internalizing problems were significantly associated with maternal mental illness (see Table 1) (r = .175, p =< .001), maternal substance abuse (r = .108, p < .001), violence exposure (r = .074, p < .001), and community adversity (r = .137, p < .001). Children's externalizing problems were significantly associated with MCJI (r = .134, p < .001), maternal mental illness (r = .142, p < .001), maternal substance abuse (r = .092, p < .001), violence exposure (r = .108, p < .001), and community adversity (r = .111, p < .001) (see Table 1).

Maternal Criminal Justice Involvement and Maternal and Environmental Risks

Children of criminal justice involved mothers were significantly more likely to be exposed to maternal substance abuse (r = .131, p < .001), community adversity (r = .102, p < .001), and violence exposure (r = .053, p < .05). There was not a significant relationship between MCJI and maternal mental illness (r = .024, p = .335) (see Table 1).

Regression Results

Internalizing Behavior Regression Model

The final model in the stepwise regression for internalizing problems accounted for 9% of the variance (F(6, 1387) = 22.9, p < .001). The regression coefficients in Table 2 illustrate that child sex and age were significantly associated with internalizing problems, such that boys and older children demonstrated elevated internalizing problems. Maternal mental illness, maternal substance abuse, and violence exposure were significantly positively associated with internalizing problems.

Externalizing Behavior Regression Model

The final model using stepwise regression for externalizing accounted for 8.3% of the variance (F(6, 1387) = 21.03, p < .001). The regression coefficients in Table 3 show that child's age was significantly related to externalizing problems, such that older children showed more externalizing behaviors than younger children. In addition, maternal mental illness, maternal substance abuse, violence exposure, and community adversity were significantly positively associated to externalizing problems. MCJI was also significantly positively associated to externalizing problems, such that children exposed to MCJI exhibited increased externalizing problems.

Moderation Regression Model

The model revealed no significant association between MCJI and internalizing problems, therefore, the moderating effects of MCJI on these variables were not analyzed. The externalizing moderation regression model accounted for 8.9% of the variance (F(9, 1384) = 15.005, p < .001). MCJI did not moderate the associations between community adversity and violence exposure and the outcome variable externalizing problems (see Table 4). MCJI had a moderating effect on the association between maternal substance abuse and externalizing problems, such that when MCJI was present, the externalizing problems of children who experienced maternal substance abuse decreased relative to when MCJI was not present. Overall, children whose mothers used substances showed the most externalizing problems, with an elevated risk for externalizing problems when their mothers were not involved with the criminal justice system.

Discussion

This study contributes to the limited knowledge base pertaining to issues affecting children assessed for maltreatment who also experience MCJI. Most of the literature that addresses dual exposure to criminal justice and child welfare systems has focused largely on children in foster care who have incarcerated parents (U.S. Government Accountability Office, 2011). Comparative study samples included children who either did not reside with their criminal justice involved caregiver, had a caregiver with a recent arrest history (for example, occurred during the child welfare investigation or six months prior), or resided with a caregiver other than their mother (Phillips et al., 2004; Phillips & Dettlaff, 2009; Phillips et al., 2009). Moreover, studies that examine the effects of MCJI used samples consisting solely of children with incarcerated mothers (Amlund-Hagen & Myers, 2003; Mignon & Ransford, 2012; Myers et al., 1999; Phillips et al., 2006; Poehlmann, 2005). The present study is unique in that the sample consisted of children who resided with their mothers during the study period and experienced MCJI during their lifetime. Findings from this study suggest that several factors beyond MCJI influence children's internalizing and externalizing problems. Nearly one-quarter of the children in this study had mothers with criminal justice involvement prior to involvement with the child welfare system. These children were more likely to encounter maternal substance abuse, violence exposure, and community adversity. Children exposed to MCJI were also more likely to exhibit higher externalizing problems than children without dual system exposure. Reasons for this association certainly vary considerably and speculation about them is beyond the scope of this paper. Nevertheless, such mental health problems can create additional familial stress, increase the risk of child maltreatment, and potentially further a mother's involvement with child welfare or criminal justice systems.

Generally, it is expected that the co-occurrence of MCJI with maternal and environmental risks would increase the likelihood of adverse childhood mental health; however, this study demonstrated that the presence of MCJI with maternal mental illness, violence exposure, and community adversity did not increase children's mental health problems. Although MCJI moderated the association between substance abuse and externalizing problems, an unexpected finding was that the presence of both risk factors had less of an impact on externalizing problems than children's exposure to maternal substance abuse alone (see Figure 1). According to Baron and Kenny (1986), a moderating variable affects the strength and direction of the relationship between the predictor and outcome. In this study, the results suggest a buffering interaction effect, such that when MCJI was present, a reduction of the magnitude of the association between substance abuse and externalizing problems occurred. A number of explanations could account for this finding. It is possible that children who experienced both MCJI and maternal substance abuse were exposed to interpersonal or social dynamics beyond the scope of what the current study captured. For example, it is possible that exposure to multiple risk factors may result in children's increased adaptability and coping. Another plausible explanation may relate to sampling, given unequal sample sizes across groups of children who experienced maternal substance abuse. Future research should continue to assess the effect of MCJI on the relationship between maternal substance abuse and childhood mental health problems.

Study Limitations

The study's results come from the NSCAW CPS sampling frame, a national child welfare population, and cannot be generalized to all children with criminal justice involved mothers. The majority of children who have experienced MCJI are not assessed for maltreatment by Child Protective Services. Therefore, generalization of these findings is most applicable to those children who live at home, whose mothers have histories of criminal justice involvement, and who have been assessed by child welfare for maltreatment. Another limitation is that the findings may be biased because of the nature of missing data. Specifically, it is possible that missingness was not completely at random on some study variables (for example, maternal substance abuse, maternal mental illness). Missing data in the regression analyses were addressed through a listwise deletion approach that resulted in a sample reduction from 1,735 to 1,381 children with complete data on all study variables. The listwise deletion approach was selected over the alternative pairwise approach because it provides a more complete case analysis on the study sample.

It was not possible to determine the onset of childhood mental pathology. The temporal effects of the introduction of an event (that is, MCJI, maternal and environmental risks) and the time participants surveyed precluded the researchers from establishing causality in this cross-sectional study. Moreover, children's internalizing and externalizing problems may have predated MCJI, maternal risks, and environmental risks. It is also difficult to discern mechanisms that result in adverse childhood mental health outcomes. For example, conditions that led to MCJI, the child's experience with MCJI, the proximal or distal timing of MCJI, and the child's developmental stage must be considered. Likewise, this difficulty in determining these mechanisms applies to buffering effects, such as that between MCJI and the association of substance abuse and children's externalizing behaviors. Moreover, the sole reliance on the mothers' assessments of mental health problems using the diagnostic CBCL measure may have resulted in mothers either over-reporting or under-reporting their children's internalizing and externalizing problems. Research on parental reports finds over-reporting and under-reporting of children's internalizing and externalizing problems. Parents' perceptions of children's behavior can be influenced by the preconceptions of what behavior they expect to see, concerns about the concepts being measured (Manders et al., 2009), or high levels of stress associated with child welfare involvement. Accessing other CBCL reporting sources (for example, teacher, child welfare caseworker) may have strengthened the validity of the study's findings. Nevertheless, research by Sheppard and Watkins (2000) on the validity of parents' assessments of child and family concerns found high levels of agreement with outside reporting sources, lending support for this study's conclusions.

Finally, the self-reported proportion of maternal substance abuse was 2.7%, lower than most child welfare national rates that range from 5% to 80%. The wide range in national rates is attributed to differences in reporting sources, definitions of substance abuse, referenced time periods, and which identified parent's behavior is reported (Loring, 2004). A potential contributor of under-reporting substance abuse in this study is that the questions on the CIDI-SF are constructed to create a composite diagnostic measure to assess substance dependence. In this study, however, individual CIDI-SF items rather than a composite score were used to determine maternal substance abuse. Similarly, the major depression scale of the CIDI-SF was the primary diagnostic assessment to measure mental illness. It is possible that mothers in this study had other types of mental health issues.

Implications for Policy and Practice

This study highlights the importance of identifying families where dual system (child welfare and criminal justice) involvement is not necessarily concurrent. Yet, neither system effectively articulates appropriate protocols for working with families with dual system involvement, nor do most policies provide clear direction on how to implement cross-system policy and practice initiatives that assess and address familial and contextual risks (U.S. General Accounting Office, 2011). The lack of coordination between the two systems creates fragmentation and gaps in services that may negatively affect families. Further, bureaucratic inertia, conflicting system goals, and lack of resources delay strategies for improved service provision. Systemwide collaboration between criminal justice and child welfare systems would require increased communication, coordination, and shared accountability for services that address the unique challenges maternal dual involvement poses for children. A first step would be for child welfare to routinely conduct comprehensive assessments, identifying mothers with current or prior involvement with the criminal justice system. Dual involvement with the two systems may require individualized interventions that focus on preventing or treating children's externalizing problems, as well as modifications to the family's social environment, where possible. Children of mothers with a history of justice involvement may also need services to resolve the potentially prolonged traumatic effects of their experience. Such provisions would similarly require that the criminal justice system screen women with children to determine the level of collaboration and coordination with child welfare systems. Implementing such practices and policies takes time, and the resources for implementation are not always readily available; nevertheless, the cost to children is too significant to delay action.

References