Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2013 May 22.
Published in final edited form as: Child Welfare. 2012;91(1):79–100.

Does Formal Integration Between Child Welfare And Behavioral Health Agencies Result in Improved Placement Stability For Adolescents Engaged With Both Systems?

Rebecca Wells 1, Emmeline Chuang 2
PMCID: PMC3661411  NIHMSID: NIHMS396126  PMID: 22894016

Abstract

National survey data were used to assess whether child welfare agency ties to behavioral health care providers improved placement stability for adolescents served by both systems. Adolescents initially at home who were later removed tended to have fewer moves when child welfare and behavioral health were in the same larger agency. Joint training of child welfare and behavioral health staff was negatively associated with numbers of moves as well as numbers of days out of home.

When children receive child protective services, caseworkers seek to ensure safety in the least disruptive manner possible while building family capacity for independence. Although the primary reason for removals from home is caregivers' inability to ensure safety, child behavior is often a contributing factor, especially for adolescents (Aarons et al., 2010; Glisson & Green, 2006). Behavioral health care for adolescents is often part of the services intended to keep families together while ensuring safety. Integration between child welfare and behavioral health providers should support family preservation through better coordination of services. The purpose of the current study was to test how ties between child welfare agencies and behavioral health care providers affected placement stability for adolescents engaged with both systems, specifically those who were able to remain with their families after maltreatment investigations had been completed. Thus, this study extends child welfare research by examining how integration with providers may help one important sub-group of children stay at home, or return more quickly if a removal does subsequently occur.

Removal of children from home has sometimes been linked to positive outcomes, especially for those at the greatest economic disadvantage (Lemmon, 2006). However, out of home placements have more frequently been linked to behavioral health problems (Doyle, 2007; Lawrence, Carlson, & Egeland, 2006; Ryan & Testa, 2005). This risk increases with subsequent moves (Ryan & Testa, 2005), especially for older children (Rubin, O'Reilly, Luan, & Localio, 2007).Recently researchers have traced the impact of placement instability as far as an increased rate of adult criminality (DeGue & Spatz Widom, 2009).

Since the passage of the Adoption Assistance and Child Welfare Act of 1980 (Public Law 96-272), child welfare agencies have made increasing efforts to keep children with their families (Frankel, 1988; McCroskey & Meezan, 1998). The primary strategy for improving family stability is to support caregivers in changing their own behavior. However, child behaviors are a significant predictor of placement changes for older children involved with child welfare (Aarons et al., 2010; Barber, Delfabbro, & Cooper, 2001). Addressing children's behavioral health may thus both directly improve their well-being and enable them to stay with their families, or at least achieve permanence more quickly after a removal from the home.

Adolescence appears to be a particularly vulnerable period to be removed from home. In a national sample of children engaged with child welfare, adolescents had a lower risk of initial out-of-home placement in the first eighteen months after engagement with child welfare than slightly younger children, but a far higher risk of removal in the subsequent eighteen months (Aarons et al., 2010) as well as greater rates of continuing placement instability (Becker, Jordan, & Larsen, 2007; Rubin et al., 2007) Research in other settings has found adolescents with behavioral problems to have elevated risk of placement breakdown (Barber et al., 2001; Oosterman, Schuengel, Wim Slot, Bullens, & Doreleijers, 2007).

The current study focused on the majority of those adolescents engaged with child welfare, who have been allowed to remain with their families after the completion of a maltreatment investigation and initiation of child protective services. In the 1999/2000 cohort of the National Survey of Child and Adolescent Well-Being used in the current study, two-thirds of adolescents receiving child welfare services two – to- six months after investigation were still in the home. Preliminary analyses indicated that these adolescents were in lower risk families than those who had been removed from their homes at that point. For instance, 17% of adolescents initially receiving in-home child welfare services had parents with mental health problems, vs. 42% of those who were out of home ( p <0.01), and 8% of those in home had a caregiver with a drug abuse problem compared to 19% for adolescents placed out of home (p<0.01). Although all children engaged with child welfare are vulnerable, remaining at home both indicates relatively low risk within this population and a key source of continuity important to continued well-being.

The core of in-home child protection is facilitation of whatever services each family needs combined with continued monitoring for progress toward independent assurance of child safety. Often such case management entails initiation of mental health care for children (Leslie et al., 2005). Such services can help adolescents with behavioral disturbances function more successfully at home (Clarke, Schaefer, Burchard, & Welkowitz, 1992) and reduce the likelihood of out of home placement(Glisson & Green, 2006).If the child is subsequently removed from home, counseling may improve placement stability by addressing by pre-existing behavioral health problems as well as the additional stressors of separation. Our interest was in both what might help adolescents at home stay with their families and, for those who were later removed, what might reduce their subsequent placement instability.

How Ties between Child Welfare and Behavioral Health Care Can Improve Outcomes for Adolescents Receiving Both Services

Child welfare agency integration with mental health providers should improve services and outcomes for adolescents who are engaged with both systems. The number of child welfare agency ties with local mental health providers has been associated in the National Survey of Child and Adolescent Well-Being (NSCAW) with greater rates of mental health service use, a stronger association between child mental health need and service use, a decreased effect of race on use, and improved child behavioral health over time (Bai, Wells, & Hillemeier, 2009; Hurlburt et al., 2004). NSCAW is the only national longitudinal survey of families engaged in the US child welfare system (Dowd et al., 2006).

In the current study, we examined associations between two key child welfare ties with providers and subsequent placement stability for adolescents receiving both child protective and behavioral health services who were initially allowed to remain in home after maltreatment investigations. First, we expected that having behavioral health care in the same agency as child protective services would improve placement stability by facilitating communication between caseworkers and counselors and reducing administrative barriers. In a field study, child welfare caseworkers spoke of walking down the hall to discuss families' mental health needs with a counselor, and noted more frequent ongoing communication with internal counselors than with those at other agencies. The mental health arm of this agency also gave priority to child protective services clients, which sometimes facilitated swifter entry into treatment. As counseling proceeds, greater ease of ongoing communication between the caseworker and counselor should enable them to support improved family functioning in a more coordinated way (Green, Rockhill, & Burrus, 2008). For instance, the caseworker and counselor could decide together when the family is ready for joint counseling, and then encourage everyone to participate. A previous NSCAW analysis found that adolescents were more likely to receive needed substance abuse treatment when this service was available in the larger agency to which child protective services belonged (Wells, Chuang, Haynes, Lee, & Bai, 2011).

Another way to facilitate case level communication is jointly training child welfare staff and behavioral health care providers. Child welfare, mental health, and substance abuse treatment staff have cited cross training as fostering better mutual knowledge and personal relationships across agencies (CSAT, 2010; Darlington & Feeney, 2008; McAlpine, Marshall, & Doran, 2001). Both group training and such individual experiences as shadowing staff at other agencies can help people understand the goals, resources, terminology, and processes each brings to their work with children and adolescents. For instance, substance abuse treatment staff have cited cross training as helpful in understanding differences between their values and those of child welfare caseworkers (CSAT, 2010). As local areas develop more advanced collaboration capacity, they become more likely to provide child welfare-behavioral health interdisciplinary training and respondents more frequently report sharing goals and plans across agencies (Drabble, 2007).

The current study builds on prior work by focusing on two key types of ties between child welfare and behavioral health care, seeking to answer the question Does formal integration between child welfare and behavioral health agencies result in improved placement stability for adolescents engaged with both systems? Based on prior findings and theory, we expected that common agency ownership and cross training would each help adolescents initially receiving in-home child protective services and behavioral health care avoid removals from home and experience fewer moves and shorter stays when removals did occur.

Method

Data

Data were drawn from the Child Protective Services (CPS) cohort of the National Survey of Child and Adolescent Well-Being (NSCAW). A two-stage stratified design was used to sample children in 92 child welfare agencies within 48 states throughout the U.S. Among those interviewed were child welfare caseworkers, child welfare agency directors, and adolescents aged 11 years or older at baseline. The first wave of interviews was conducted 2 – 6 months after the initial child welfare investigation or assessment, between October 1999 and December 2000. The current study also uses three additional waves of data, collected at 12 months (Wave 2), 18 months (Wave 3), and 36 months (Wave 4) after the initial investigation.

Caseworkers were interviewed about child placements in all four waves. Child welfare agency directors were interviewed at Wave 1 only, with questions addressing ties to providers as well as other agency level factors. Adolescents 11 years and older were interviewed about behavioral health service receipt at Wave 1, Wave 3, and Wave 4. For the present analyses, Research Triangle Institute (RTI) International also merged 2000Area Resource File data with NSCAW data on local agencies using county-level identifiers. The original data collection was approved by RTI International's IRB. Analyses for the current study were approved by the lead author's institutional review board.

Sample

Of the 5,501 children within the full NSCAW sample who were investigated for child maltreatment, 4,080 received some services from a child welfare agency, including case management. The current study focused on the subset of these children who were 11 years or older and still located in home 2-6 months after the completion of a maltreatment investigation(Wave 1), who also received behavioral health services at some point in the next 36 months (Waves 1 – 4). A total of 901adolescents were 11 years or older at baseline and 600 were still at home. Over half (349)of those 600 reported receiving counseling from a school counselor, doctor, or therapist for feeling depressed and/or professional treatment for an alcohol or drug problem within 36 months after entering the child welfare system. Of these 349 adolescents, 275 were served by child welfare agencies whose directors were also interviewed as part of NSCAW. Only public child welfare agencies were included in NSCAW (DHHS, 2001).

List wise deletion for item missingness reduced the final analytic sample from 275 to 246adolescentslocated within 71 child welfare agencies in the model predicting any out-of-home placements after baseline. Using IVEware in SAS statistical software, multiple imputation was attempted to maintain the full sample size. However, high imputation-to-imputation variance in estimated values for inter-agency ties indicated an inability to impute these values reliably, despite incorporation of extensive agency-level data in the information matrix (Raghunathan, Solenberger, & Hoewyk, 2002). Therefore, unimputed data were used in the regression analyses.

The sampling weights within NSCAW adjust for survey non-response but not item non-response (Dowd et al., 2006; Little, 1988). T-tests conducted to determine if the data could be considered missing at random revealed that adolescents omitted from the sample due to missing data averaged 8 months younger than those in the sample.

Models examining adolescents' total number of placements and total days spent out-of-home were further restricted to those with at least one out-of-home placement. A total of 118adolescents met this criterion. List wise deletion reduced these samples to 88 adolescents within 48 agencies.

Measures

Placement stability

Three variables were used to measure adolescent placement stability: (1) whether the adolescent was removed from home at any point in the 36 months after beginning in-home child protective services; (2) the total number of out-of-home placements experienced in those 36 months; and (3) the cumulative number of days spent out-of-home during the same time period.

Child welfare agency ties with behavioral health care providers

The first measure was a 1/0 binary indicator of whether or not mental health and/or substance abuse treatment were in the same agency as child protective services. The second measure was a separate 1/0 binary indicator of child welfare agency cross training of staff with mental health and/or substance abuse treatment providers. Both measures were from the interview with the child welfare agency director.

Other covariates

A number of measures were included to account for potential confounders of the associations between inter-agency ties and placement stability. The first variable, drawn from Area Resource File data, indicated whether the agency was located within a non-metropolitan statistical area, and served as a control for rural-urban differences in service provision as well as a proxy for organizational size (Belanger & Stone, 2008).

Child and family attributes that might influence placement stability were also included as covariates. These factors were adolescent age (Wulczyn, Kogan, & Harden, 2003); gender (McMahon & Clay-Warner, 2002; Smith, Stormshak, Chamberlain, & Whaley, 2001); race; type of maltreatment identified through investigation as the most serious in each case (Bundy-Fazioli, Winokur, & DeLong-Hamilton, 2009; Garland, Landsverk, Hough, & Ellis-MacLeod, 1996); standardized Child Behavior Checklist (CBCL) externalizing and internalizing scores indicating behavioral health problems (Achenbach & Edelbrock, 1991; Becker et al., 2007; James, 2004); and a three-level categorical measure of family risk based on a range of factors, including difficulty providing necessities, prior history of abuse or neglect, and domestic violence (Farmer, Southerland, Mustillo, & Burns, 2009).

Analytic Plan

The NSAW sample included multiple adolescents within each child welfare agency. Results of fully unconditional random intercept models indicated significant variance across agencies in adolescent placement instability. However, the final analytic samples contained modest numbers of units of higher aggregation (48-71 child welfare agencies). With unbalanced data such as NSCAW, estimation of coefficients and standard errors in multilevel models rely on large-sample theory. With binary and count dependent variables, a small sample of level-2 units (<100) can result in biased estimates premised on inaccurate assumptions about variable distributions (Raudenbush & Bryk, 2002). Therefore, all models were analyzed as single-level models, with the adjustments described below to accommodate correlations of adolescent outcomes within agencies.

Next, descriptive statistics were generated using Stata 10′s -svy- module, which incorporated the sample's stratification, clustering of adolescents within agencies (the primary sampling units), and probability weights, and thus yields statistics representative of children engaged with child welfare throughout most of the United States (P Biemer, Christ, Wheeless, & Wiesen, 2006; Paul Biemer, Christ, Wheeless, & Wiesen, 2008; Bollen, 1989; StataCorp, 2007; L. StataCorp, 2007). Pearson and tetrachoric correlations were used to assess possible bivariate collinearity between measures in both final analytic samples (Bollen, 1989), and tolerances were used to test for multicollinearity (Hamilton, 1992).

Multiple regression analyses were also conducted using the Stata 10.0 -svy- module, thus again adjusting estimates to accommodate the complex survey design. The link function used for each model reflected the nature of the dependent variable: logistic for whether adolescents experienced an out-of-home placement and negative binomial for the numbers of placement moves and days spent out-of-home. The model predicting any out of home placement was estimated for all adolescents receiving behavioral health care who were in home at baseline. The models of numbers of out-of-home placements and days out of home were estimated only for adolescents who were removed from home after initially receiving in-home child protection services.

Statistical power analyses were conducted using Optimal Design to accommodate the clustering of adolescents within agencies (Spybrook, Raudenbush, Liu, Congdon, & Martínez, 2008). Results for the model predicting any out-of-home placement indicated an 80% likelihood of detecting medium effect sizes for behavioral health care within the larger agency and for joint training in the full sample at alpha=0.10. In the models predicting placement instability for adolescents who had been removed from home, only large effect sizes could be detected. This meant that non-significant results would be ambiguous for these models. Because of the limited statistical power, we set the threshold for statistical significance in all models at alpha=0.10.

Results

Descriptive Statistics

Table 1 provides weighted descriptive statistics for all adolescents initially at home who received some type of behavioral health care within 36 months of beginning in-home child protective services (n=246) as well as for the subset of 88 who were later removed from home.

Table 1. Weighted Descriptive Statistics.

Full Sample Later Removed from Home
n=246 n=88
Mean/ % SE 95% CI Mean/ % SE 95% CI
 Any mental health counseling 92% 88% 96% 94% 87% 100%
 Any substance abuse treatment 33% 25% 42% 45% 29% 61%
 Only mental health counseling 66% 58% 75% 55% 39% 71%
 Only substance abuse counseling 8% 4% 12% 6% 0% 13%
 Mental health and substance abuse counseling 25% 16% 34% 39% 22% 55%
Removed from home after baseline 28% 21% 35%
Number of out-of-home placements (given any) 2.75 0.26 2.24 3.27
Number of days out of home (given >=1 removal) 452.32 42.75 367.30 537.34
 Mental health in larger agency 12% 4% 20% 14% 6% 1% 26%
 Substance abuse tmt in larger agency 12% 4% 20% 15% 6% 3% 28%
Any behavioral health in larger agency 14% 6% 23% 18% 7% 4% 31%
 Cross training with mental health providers 50% 31% 69% 49% 10% 28% 69%
 Cross training with substance abuse tmt 49% 30% 68% 46% 10% 26% 66%
Cross training with any behavioral health providers 62% 43% 80% 58% 10% 38% 78%
Agency in a non-metropolitan area 13% 3% 22% 15% 7% 2% 29%
Child age at baseline (years) 12.69 0.12 12.44 12.93 12.76 0.21 12.34 13.18
Child was a boy 48% 37% 59% 37% 7% 24% 51%
Child was white 39% 27% 52% 38% 9% 21% 55%
Most serious maltreatment: physical 29% 24% 34% 37% 7% 24% 50%
Most serious maltreatment: sexual 12% 6% 18% 4% 2% -1% 9%
Externalizing score 62.91 1.27 60.37 65.44 64.79 2.00 60.82 68.77
Internalizing score 58.60 1.26 56.09 61.10 59.05 2.50 54.08 64.02
Composite measure of family risk: low 28% 20%
Composite measure of family risk: medium 38% 48%
Composite measure of family risk: high 33% 33%
Open in a new tab

Table 1 begins with how often the adolescents in each sample received some type of mental health counseling (92% in the overall sample and 94% among those later removed from home), substance abuse treatment (33% overall and 45% among those later removed), only mental health counseling (66%vs. 55%), only substance abuse treatment (8%vs. 6%), or both (25% in the initial sample and 39% among those later removed from home). Approximately 28% of adolescents in home at baseline were subsequently placed out-of-home. On average, that latter group experienced 2.75 out-of-home placements and spent a total of 452 days out-of-home.

Child welfare agencies with ties to mental health providers tended to have ties with substance abuse treatment providers as well. About one in seven (14%) of agencies in the initial sample were in the same agency as a mental health and/or substance abuse treatment provider. Almost two thirds of child welfare agency heads reported cross-training with mental health and/or substance abuse treatment providers (62%). About one in eight (13%) of child welfare agencies were in non-metropolitan areas.

The average age at baseline was just under 13 years. Roughly half (48%) of the adolescents in the sample were male and 39% were white. Physical abuse was identified as the most serious form of maltreatment in 29% of the cases and sexual abuse in 12% of the cases. The mean standardized Child Behavioral Checklist externalizing score was 62.91 and the mean internalizing score was 58.60 at wave 1. Roughly one third of the adolescents' families were in each risk category based on a range of factors (28% “low,” 38% “medium,” and 33% “high”). The highest bivariate correlation was 0.58, between externalizing and internalizing CBCL scores in the full final sample of adolescents (n=246); this correlation was 0.49 in the sample restricted to adolescents removed from home after baseline. Although this may have reduced the precision with which the unique effects of each dimension of behavioral problems affected placement stability, these correlations had no effect on estimation of associations between child welfare agency ties and placement stability (Bollen, 1989). The highest bivariate correlation between either key predictor and any other predictor was -0.40 between agency cross-training and children's externalizing symptom in the full final sample. The lowest tolerance for either type of child welfare agency tie in either sample was 0.87, also indicating the absence of multicollinearity (Hamilton, 1992).

Final Model Results

Table 2 shows the multiple regression results. Neither common agency ownership nor cross training between child welfare and behavioral health care providers was associated with the likelihood that adolescents would experience an out of home placement after baseline. Common agency ownership was associated with 19% fewer placement changes among adolescents who were later removed (IRR=0.81, p<0.10). Cross training was associated with 36% fewer out-of-home moves (IRR=0.64, p<0.01) and 42% fewer days out of home (IRR=0.58, p<0.01).

Table 2. Regression Models.

For Adolescents in Home at Baseline: For Adolescents Removed from Home After Baseline:
Subsequent Out-of-Home Placement Number of Out-of-Home Moves Number of Days Out of Home
Logistic (n=246) Negative Binomial (n=88) Negative Binomial (n=88)
OR SE P-value 95% CI IRR SE P-value 95% CI IRR SE P-value 95% CI
Behavioral health in larger agency 1.35 0.61 0.55 3.29 0.81 0.10 + 0.64 1.04 0.75 0.19 0.45 1.25
Cross training with behavioral health 0.6 0.24 0.27 1.35 0.64 0.08 ** 0.50 0.82 0.58 0.11 ** 0.41 0.84
Agency in a non- metropolitan area 1.75 0.98 0.57 5.34 1.22 0.22 0.85 1.73 2.05 0.52 ** 1.24 3.38
Child age at baseline (years) 0.97 0.16 0.70 1.34 0.95 0.05 0.86 1.05 1 0.07 0.87 1.15
Child was a boy 0.38 0.22 0.12 1.22 0.87 0.11 0.68 1.11 0.55 0.12 ** 0.36 0.83
Child race white 1.11 0.54 0.42 2.94 0.58 0.07 *** 0.45 0.75 0.39 0.07 *** 0.27 0.56
Most serious maltreatment: physical 1.38 0.55 0.62 3.07 1.11 0.12 0.89 1.38 1.23 0.24 0.84 1.8
Most serious maltreatment: sexual 0.16 0.11 ** 0.04 0.60 1.12 0.21 0.77 1.62 0.91 0.22 0.56 1.47
Externalizing score 1.03 0.02 0.99 1.09 1.01 0.01 1.00 1.02 0.98 0.01 * 0.97 1
Internalizing score 0.98 0.02 0.94 1.03 0.99 0.01 0.98 1.01 1.01 0.01 * 1 1.03
Composite measure of risk 1.38 0.28 0.92 2.07 0.95 0.07 0.82 1.11 1.5 0.25 * 1.08 2.07
Probability > F statistic 0.12 0.00 0.00
Open in a new tab
+

<0.10

*

<0.05

**

<0.01

***

<0.001

There were several associations between control variables and placement stability as well. Boys tended to experience shorter durations out-of-home (IRR 0.55, p<0.01). White adolescents experienced fewer placement changes (IRR=0.58, p<0.01) and shorter out of home stays (IRR=0.39, p<0.01) than non-white adolescents. Compared to other types of maltreatment, sexual abuse was negatively associated with an initial removal from home (OR 0.16, p<0.01).Externalizing problems were associated with shorter out-of-home stays (IRR=0.98, p<0.05), whereas internalizing problems were associated with shorter out-of-home stays (IRR=1.01, p<0.05). Adolescents from higher risk families tended to remain out of the home longer (IRR=1.5, p<0.05).

Discussion

The ties between child welfare and behavioral health care providers examined in this study do not appear to prevent removal from home for adolescents who begin in home, but may reduce placement instability for those who are later moved into out-of-home settings. First, common agency ownership appeared to reduce the number of placement changes for adolescents receiving behavioral health care who were removed from home at some point after the first several months of engagement with child welfare. This finding is in keeping with a recent finding from the same survey that adolescents were more likely to receive substance abuse treatment when this service child protective services were available within the same agency (Wells et al., 2011).

The majority of local child welfare agencies in the NSCAW sample were units of larger agencies, and among those, substance abuse treatment and mental health were also frequently offered (DHHS, 2001). Findings from the current study indicate that the recent trend toward omnibus health and human service agencies (DHHS, 2001) may serve families well. However, in some areas adding behavioral health care to public agencies may be either infeasible or else undesirable because of strong community-based providers available locally. It may be difficult for state-administered child welfare systems to take such variation into account in restructuring local agencies. Finally, consolidation may actually reduce service capacity if used as an opportunity to reduce costs.

Different external stakeholders, laws and regulations, and funding streams for child welfare, mental health, and substance abuse treatment also complicate provision of these services within a single organization. For instance, the Children's Bureau oversees child protective services and audits selected local child welfare agencies for progress on performance goals. Class action lawsuits have led to court oversight of child protective services in many states (Kosanovich & Joseph, 2005). Medicaid and other public insurance and block grants pay for – and audit – mental health and substance abuse treatment providers. Local courts referring individuals into substance abuse treatment have different goals and timelines than child welfare agencies. Thus, although child safety and behavioral health needs are closely intertwined for many of the individuals served, the disparate accountability requirements and funding attendant to provision of different services requires substantial leadership capacity and infrastructure to address within common organizations.

Providing both child welfare and behavioral health services within a common agency may be easier in nonprofits than for public agencies. Although state child welfare systems operate under a federal mandate to provide child protective services, in all states at least some services are provided by private agencies through purchase service contracts (Collins-Camargo, Ensign, & Flaherty, 2008). In-home services are the most commonly contracted out portion of child welfare services (DHHS, 2001). As illustrated by the current sample, families receiving in-home services often also receive some type of behavioral health care. Some of the nonprofits most active in child protective services, such as those under religious auspices, have family-centered missions that make provision of multiple health and social services compatible with existing organizational values. Nonprofits are also buffered from local politics because they report to boards of directors who often oversee entire states or regions. Nonprofits may therefore be able to decide how many services to offer based on a range of relevant factors, including the depth of management capacity available to manage the attendant complexity and prior related organizational experience. Current research on private child and family serving agencies suggests that they tend to be large, mature organizations with diverse service arrays including behavioral health services in addition to child welfare (McBeath, Collins-Camargo, & Chuang, 2011). Thus, when common agency ownership is not feasible for public agencies, similar benefits may be achieved by contracting with nonprofits that provide multiple services.

Even when common agency ownership is not feasible at all, co-location may improve service use for individuals engaged with more than one system. For the individuals themselves, having more than one service in a single place can simplify transportation logistics and thus greatly improve access (Azzi-Lessing & Olsen, 1996). For the professionals, co-location could facilitate more complete and timely information sharing. Memoranda of understanding may outline processes for sharing information about individuals between agencies. When co-location of entire agencies is not feasible, basing staff from one agency in another can achieve some of the same benefits (Lee, Esaki, & Greene, 2009).

Although cross training between child welfare agencies and behavioral health care providers did not prevent removals from home, cross training was associated with fewer placement changes and shorter out-of-home stays for adolescents removed from home after initially receiving in-home child protective services. Cross training creates the opportunity for front line staff to develop better mutual understanding and personal relationships. Such relationships are frequently cited by staff as critical to facilitating effective coordination for the families they serve (CSAT, 2010; Darlington & Feeney, 2008; McAlpine et al., 2001).

As communication technologies improve, training on how to work across agencies is increasingly available online (e.g., http://www.nccollaborative.org/). Online training can lower costs for agencies and improve flexibility for staff. However, this mode of training currently loses the opportunity for staff to interact with people from other agencies. Corporations and the federal government are increasingly using audiovisual conferencing technologies that replicate much of the richness of face-to-face interaction for entire groups. Given high equipment costs, it seems unlikely that many local health and human service providers will have access to these capabilities in the near future. However, given the relative proximity of many agency staff within any given service area, it may be feasible to include more than one agency at a common location and connect at least one additional group at a more distant site through an inexpensive technology such as Skype or Adobe Connect.

Managers might maintain some balance of face-to-face and remote training in order to foster strong relationships at affordable cost. For instance, face-to-face training might be conducted annually, with time allowed for people to socialize. In addition, all new workers could complete the online training within the first few months of employment. This would ensure that everyone has some exposure to basic information about other agencies serving their families, as well as periodic opportunities to build essential personal relationships. Providing ongoing training requires sustainable funding for both the trainers and those undergoing training. Ideally, agencies will build shared training into budgets every year. Policy makers can support sustainability by making joint training count toward requirements for each participating agency, e.g., both evidence of progress toward child welfare goals and compliance with Medicaid service definitions for behavioral health care providers.

Two of the coefficients for covariates were surprising. First, adolescents whose most serious maltreatment was identified as sexual abuse were less likely to be subsequently removed from in-home after initially receiving in-home services. Sexually abused children may generally have been allowed to remain in home only when perpetrators were gone. Thus is it possible that at that point those children were generally considered at relatively low risk of repeated maltreatment. The second surprising covariate was the negative association between externalizing behaviors and the number of days out of home, given a removal. Conversely, and more intuitively, higher internalizing scores were associated with longer durations out of home.

Some aspects of the available data limit ability to draw inferences. Some of the adolescents undoubtedly received behavioral services from providers besides those to which the child welfare agency had ties. Although the data were from a national sample and probability weights adjusted for survey non-response, the samples examined here were also small, raising the possibility that other groups of adolescents would have different experiences. In addition, the limited statistical power to detect small-to-medium effect sizes rendered non-significant results ambiguous. Unfortunately, the authors are not aware of any forthcoming surveys that will yield larger samples of child welfare agencies, and thus greater statistical power to detect the effects of agency level strategies.

Implications for Practice

This is the first study to trace child welfare agency connections with behavioral health care to placement stability. Results suggest that adolescents receiving both in-home child protective and behavioral health services benefit from common agency ownership as well as cross training. More broadly, we believe the current study findings indicate that families benefit when professionals addressing different facets of their lives have opportunities to develop mutual understanding and maintain ongoing coordination. Significant obstacles impede such connections between front line staff, including and managerial and sometimes political complications attendant to increasing the range of services provided by any one agency, as well as high workloads for staff, their supervisors, and managers; high turnover; and lack of reimbursement for provider time devoted to coordination(Darlington & Feeney, 2008; Green et al., 2008). Nonetheless, we hope that policy makers as well as agency administrators and front line supervisors will capitalize on any opportunities to promote personal relationships between child welfare staff and behavioral health care providers. In addition to formal agency integration, other possibilities may include participation in learning collaboratives through which staff from multiple agencies meet repeatedly to learn new practices, and making introductions to staff at other agencies part of the standard initiation process for new employees.

Keeping children safely at home with their parents is a top child welfare priority. Child welfare agency integration with behavioral health providers serving their families may help achieve this goal. We hope health and human service leadership will continue to explore what types of integration can best facilitate these and other aspects of child well-being.

Acknowledgments

This work was supported through National Institute of Mental Health grant # 5K01MH076175. This study used data from the National Survey on Child and Adolescent Well-Being, which was developed under contract with the Administration on Children, Youth, and Families, U.S. Department of Health and Human Services. The data have been provided by the National Data Archive on Child Abuse and Neglect. The information and opinions expressed herein reflect solely the position of the author(s).

Contributor Information

Rebecca Wells, Email: rwells@unc.edu, Department of Health Policy and Management, University of North Carolina, McGavran-Greenberg CB# 7411, Chapel Hill, NC 27599-7411, Phone 919-966-7384, Fax 919-966-6961.

Emmeline Chuang, Email: echuang@mail.sdsu.edu, Division of Health Management and Policy, Graduate School of Public Health, San Diego State University, San Diego, CA 92182-4162, Hardy Tower 169, (619)-594-6439.

References

  1. Aarons GA, James S, Monn AR, Raghavan R, Wells RS, Leslie LK. Behavior problems and placement change in a national child welfare sample: a prospective study. Journal of the American Academy of Child & Adolescent Psychiatry. 2010;49(1):70–80. doi: 10.1097/00004583-201001000-00011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Achenbach TM, Edelbrock C. Manual for the Child Behavior Checklist and 1991 Profile. Burlington, VT: University of Vermont Department of Psychiatry; 1991. [Google Scholar]
  3. Azzi-Lessing L, Olsen LJ. Substance Abuse-Affected Families in the Child Welfare System: New Challenges, New Alliances. Social Work. 1996;41(1):15–23. doi: 10.1093/sw/41.1.15. [DOI] [PubMed] [Google Scholar]
  4. Bai Yu, Wells Rebecca, Hillemeier Marianne M. Coordination between child welfare agencies and mental health service providers, children's service use, and outcomes. Child Abuse & Neglect. 2009;33(6):372–381. doi: 10.1016/j.chiabu.2008.10.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Barber JG, Delfabbro PH, Cooper LL. The predictors of unsuccessful transition to foster care. The Journal of Child Psychology and Psychiatry and Allied Disciplines. 2001;42(06):785–790. [PubMed] [Google Scholar]
  6. Becker MA, Jordan N, Larsen R. Predictors of successful permanency planning and length of stay in foster care: The role of race, diagnosis and place of residence. Children and Youth Services Review. 2007;29(8):1102–1113. [Google Scholar]
  7. Belanger K, Stone W. The social service divide: Service availability and accessibility in rural versus urban counties and impact on child welfare outcomes. Child welfare. 2008;87(4):101–124. [PubMed] [Google Scholar]
  8. Biemer P, Christ S, Wheeless S, Wiesen C. National Survey of Child and Adolescent Well-Being (NSCAW): Statistical user's manual. New York: National Data Archive on Child Abuse and Neglect; 2006. [Google Scholar]
  9. Biemer Paul, Christ Sharon, Wheeless Sara, Wiesen Chris. Statistical User's Manual: Research Triangle Institute. University of North Carolina at Chapel Hill, Caliber Associates, Children's Hospital; San Diego: 2008. National Survey of Child and Adolescent Well-Being NSCAW. [Google Scholar]
  10. Bollen KA. Structural equation models with latent variables. New York 1989 [Google Scholar]
  11. Bundy-Fazioli K, Winokur M, DeLong-Hamilton T. Placement outcomes for children removed for neglect. Child welfare. 2009;88(3):85–102. [PubMed] [Google Scholar]
  12. Clarke RT, Schaefer M, Burchard JD, Welkowitz JW. Wrapping community-based mental health services around children with a severe behavioral disorder: An evaluation of Project Wraparound. Journal of Child and Family Studies. 1992;1(3):241–261. [Google Scholar]
  13. Collins-Camargo C, Ensign K, Flaherty C. The National Quality Improvement Center on the Privatization of Child Welfare Services: A Program Description. Research on Social Work Practice. 2008;18(1):72. [Google Scholar]
  14. CSAT. Substance Abuse Specialists in Child Welfare Agencies and Dependency Courts Considerations for Program Designers and Evaluators. Rockville, MD: Substance Abuse and Mental Health Services Administration; 2010. HHS Pub No (SMA) 10-4557. [Google Scholar]
  15. Darlington Y, Feeney JA. Collaboration between mental health and child protection services: Professionals' perceptions of best practice. Children and Youth Services Review. 2008;30(2):187–198. [Google Scholar]
  16. DeGue S, Spatz Widom C. Does out-of-home placement mediate the relationship between child maltreatment and adult criminality? Child Maltreatment. 2009;14(4):344. doi: 10.1177/1077559509332264. [DOI] [PubMed] [Google Scholar]
  17. DHHS. National Survey of Child and Adolescent Well-Being: Local child welfare agency survey. Washington, DC: U.S. Department of Health and Human Services, Administration for Children, Youth and Families; 2001. Report. [Google Scholar]
  18. Dowd K, Kinsey S, Wheeless S, Thissen RJ, Richardson J, Suresh R, Lytle T. National Survey of Child and Adolescent Well-Being (NSCAW) - Combined waves 1-4 data file user's manual restricted release version. Durham, NC: Research Triangle Institute; 2006. [Google Scholar]
  19. Doyle Joseph J. Child Protection and Child Outcomes: Measuring the Effects of Foster Care. [Article] American Economic Review. 2007;97(5):1583–1610. doi: 10.1257/aer.97.5.1583. [DOI] [PubMed] [Google Scholar]
  20. Drabble L. Pathways to Collaboration: Exploring Values and Collaborative Practice Between Child Welfare and Substance Abuse Treatment Fields. Child Maltreatment. 2007;12(1):31. doi: 10.1177/1077559506296721. [DOI] [PubMed] [Google Scholar]
  21. Farmer EM, Southerland D, Mustillo SA, Burns BJ. Returning home in systems of care: Rates, predictors, and stability. Journal of Emotional and Behavioral Disorders. 2009;17(3):133–146. [Google Scholar]
  22. Frankel H. Family-centered, home-based services in child protection: A review of the research. Social Service Review. 1988;62(1):137–157. [Google Scholar]
  23. Garland AF, Landsverk J, Hough RL, Ellis-MacLeod E. Type of maltreatment as a predictor of mental health service use for children in foster care. Child abuse & neglect. 1996;20(8):675–688. doi: 10.1016/0145-2134(96)00056-7. [DOI] [PubMed] [Google Scholar]
  24. Glisson C, Green P. The role of specialty mental health care in predicting child welfare and juvenile justice out-of-home placements. Research on Social Work Practice. 2006;16(5):480. [Google Scholar]
  25. Green BL, Rockhill A, Burrus S. The Role of Interagency Collaboration for Substance-Abusing Families Involved with Child Welfare. Child Welfare. 2008;87(1):33. [PubMed] [Google Scholar]
  26. Hamilton Lawrence C. Regression with graphics: A second applied course in statistics. Belmont, California: Duxbury Press; 1992. [Google Scholar]
  27. Hurlburt Michael S, Leslie Laurel K, Landsverk John, Barth Richard P, Burns Barbara J, Gibbons Robert D, et al. Zhang Jinjin. Contextual predictors of mental health service use among children open to child welfare. Archives of General Psychiatry. 2004;61(12):1217–1224. doi: 10.1001/archpsyc.61.12.1217. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. James Sigrid. Why Do Foster Care Placements Disrupt? An Investigation of Reasons for Placement Change in Foster Care. The Social Service Review. 2004;78(4):601–627. [Google Scholar]
  29. Kosanovich Amy, Joseph Rachel Molly. Child welfare consent decrees: Analysis of thirty-five court actions from 1995 to 2005. Washington DC: Child Welfare League of America; 2005. p. 56. [Google Scholar]
  30. Lawrence CR, Carlson EA, Egeland B. The impact of foster care on development. Development and Psychopathology. 2006;18(01):57–76. doi: 10.1017/S0954579406060044. [DOI] [PubMed] [Google Scholar]
  31. Lee E, Esaki N, Greene R. Collocation: Integrating child welfare and substance abuse services. Journal of Social Work Practice in the Addictions. 2009;9(1):55–70. [Google Scholar]
  32. Lemmon John H. The Effects of Maltreatment Recurrence and Child Welfare Services on Dimensions of Delinquency. Criminal Justice Review. 2006;31(1):5–32. doi: 10.1177/0734016806287945. [DOI] [Google Scholar]
  33. Leslie LK, Hurlburt MS, James S, Landsverk J, Slymen DJ, Zhang J. Relationship between entry into child welfare and mental health service use. Psychiatr Serv. 2005;56(8):981–987. doi: 10.1176/appi.ps.56.8.981. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Little Roderick JA. Missing data in large surveys. Journal of Business and Economic Statistics. 1988;6:287–301. [Google Scholar]
  35. McAlpine C, Marshall CC, Doran NH. Combining child welfare and substance abuse services: A blended model of intervention. Child Welfare. 2001;80(2):129–149. [PubMed] [Google Scholar]
  36. McBeath B, Collins-Camargo C, Chuang E. National Quality Improvement Center on the Privatization of Child Welfare Services 2011 [Google Scholar]
  37. McCroskey J, Meezan W. Family-centered services: Approaches and effectiveness. The Future of Children. 1998;8(1):54–71. [PubMed] [Google Scholar]
  38. McMahon J, Clay-Warner J. Child Abuse and Future Criminality. Journal of Interpersonal Violence. 2002;17(9):1002. [Google Scholar]
  39. Oosterman Mirjam, Schuengel Carlo, Wim Slot N, Bullens Ruud AR, Doreleijers Theo AH. Disruptions in foster care: A review and meta-analysis. Children and Youth Services Review. 2007;29(1):53–76. doi: 10.1016/j.childyouth.2006.07.003. [DOI] [Google Scholar]
  40. Raghunathan TE, Solenberger PW, Hoewyk JV. IVEware: Imputation and variance estimation software user guide. Ann Arbor, Michigan: Survey Methodology Program at the Survey Research Center Institute for Social Research at the University of Michigan; 2002. [Google Scholar]
  41. Raudenbush SW, Bryk AS. Hierarchical linear models: Applications and data analysis methods. Thousand Oaks, CA: Sage; 2002. [Google Scholar]
  42. Rubin DM, O'Reilly ALR, Luan X, Localio AR. The impact of placement stability on behavioral well-being for children in foster care. Pediatrics. 2007;119(2):336. doi: 10.1542/peds.2006-1995. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Ryan JP, Testa MF. Child maltreatment and juvenile delinquency: Investigating the role of placement and placement instability. Children and Youth Services Review. 2005;27(3):227–249. [Google Scholar]
  44. Smith DK, Stormshak E, Chamberlain P, Whaley RB. Placement disruption in treatment foster care. Journal of Emotional and Behavioral Disorders. 2001;9(3):200–205. [Google Scholar]
  45. StataCorp. Stata Statistical Software: Release 10. College Station, TX: Stata Corporation LP; 2007. [Google Scholar]
  46. StataCorp LP. Stata statistical software: Release 10.0. College Station, TX: Stata Corporation; 2007. [Google Scholar]
  47. Wells Rebecca, Chuang Emmeline, Haynes Lindsey E, Lee I Heng, Bai Yu. Child welfare agency ties to providers and schools and substance abuse treatment use by adolescents. Journal of Substance Abuse Treatment. 2011;40(1):26–34. doi: 10.1016/j.jsat.2010.08.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Wulczyn FW, Kogan JN, Harden BJ. Placement stability and movement trajectories. Social Service Review. 2003;77(2):212–236. [Google Scholar]

RESOURCES