Foster care assessment: A study of the placement decision process in Flanders

G Model CHIABU-2964; No. of Pages 9

ARTICLE IN PRESS Child Abuse & Neglect xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Child Abuse & Neglect

Foster care assessment: A study of the placement decision process in Flanders Johan Vanderfaeillie a,∗ , Huub Pijnenburg b , Harm Damen c , Frank Van Holen a,d a b c d

Department of Clinical and Lifespan Psychology, Vrije Universiteit Brussel, Brussels, Belgium HAN University of Applied Sciences, Praktikon and Radboud University, Nijmegen, The Netherlands Praktikon, Nijmegen and Pactum, Arnhem, The Netherlands Pleegzorg Vlaams-Brabant en Brussel, Brussels, Belgium

a r t i c l e

i n f o

Article history: Received 15 October 2014 Received in revised form 15 March 2015 Accepted 5 April 2015 Available online xxx Keywords: Child welfare Children in need (services for) Foster care (Family) Placement decision

a b s t r a c t Family foster care placement decision-making has a weak scientific underpinning. Mostly a ‘variable-oriented approach’ is taken, which requires a lot of information that is not always available. The identification of clusters of foster children with similar characteristics may be a more viable decision strategy. In this study we investigated if foster children could indeed be clustered, which problems were identified at the time of placement, and the influence of placement history. It proved possible to group foster children into two clusters: (1) young children with familial problems and few behavioral problems, and (2) older children with prominent child problems and behavioral problems. For foster children with and without placement history, problems associated with placement proved identical. Considering that a foster care placement did not result in fundamental change in the problems present at time of placement, the importance is stressed of approaching foster care assessment as part of a decision making process which looks back as well as forward. Placement decisions should be based on an appraisal of the appropriateness of foster placement as a solution for the child. In conjunction with this appraisal a decision is required on how parents can be supported toward reunification. Or – if this is not an option – whether long term foster care is the best option for the child and if so, what conditions need to be met. © 2015 Elsevier Ltd. All rights reserved.

Introduction A family foster care placement is the option of choice for children in need of out-of-home care, because foster families provide more continuity in relationships than for example a group home. Compared to children placed in institutions, foster children are more likely to grow up into well-functioning adults (Barber, Delfabbro, & Cooper, 2001; Strijker & Zandberg, 2001; Van IJzendoorn, 2010). Family foster care is however a vulnerable intervention and questions can be raised as to its efficacy. Evidence for this can be found in the high number of breakdowns (premature terminations of the foster placement for negative reasons), the increase of behavioral problems among foster children during their stay in foster care, and the relatively unfavorable image of alumni foster children leaving foster care (Hermanns, 2010; Vanderfaeillie & Van Holen, 2010; Vanderfaeillie, Van Holen, Vanschoonlandt, Robberechts, & Stroobants, 2013). These problems may result from placing children in family foster care, for whom this intervention is not adequate (Vanderfaeillie, Van Holen, & Coussens, 2008). This often occurs when the assessment preceding the foster care placement was not carried out carefully.

∗ Corresponding author address: Vrije Universiteit Brussel, Department of Clinical and Lifespan Psychology, Pleinlaan 2, 1050 Brussels. http://dx.doi.org/10.1016/j.chiabu.2015.04.002 0145-2134/© 2015 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Vanderfaeillie, J., et al. Foster care assessment: A study of the placement decision process in Flanders. Child Abuse & Neglect (2015), http://dx.doi.org/10.1016/j.chiabu.2015.04.002

G Model CHIABU-2964; No. of Pages 9

2

ARTICLE IN PRESS J. Vanderfaeillie et al. / Child Abuse & Neglect xxx (2015) xxx–xxx

Foster children are a heterogeneous group having only one thing in common: they do not live with their parents (Enzlin, 1996). It is obvious that by placing only minors who are well suited for foster care, the risk of a placement breakdown can be minimized. Such a strategy would however be at odds with the current policy of considering foster care as the preferred option when of out of home placement is required. Increasing our understanding of problems (parenting, familial and child problems) impacting the likelihood of a family foster care placement can result in more effective placement decisions. If family foster care is to fulfill its increasingly important role in youth care today successfully, it must become clearer in which cases this important child welfare intervention is promising. The aim of this study is to increase our understanding of the interrelation of child characteristics and (parenting, familial and child) problems at the time of placement. Research into clusters of children in foster care For birth parents and child alike placing a child out-of-home is a far-reaching decision. As in other countries, in Flanders out-of-home placement only comes into view when the upbringing of a child in its birth family is endangered and/or a child’s safety or development (physical, affective, moral, intellectual or social) are threatened by relational conflicts and/or living conditions. The main reasons for placing a child in out-of-home care are parenting and familial problems, and behavioral and emotional problems in the child. Some authors argue that parenting problems (e.g. neglect and abuse) and familial problems (e.g. inadequate housing, poverty and domestic violence) carry more weight in placement decisions than do children’s behavioral problems. In 75–80% of out-of-home placements parenting and familial problems are an issue (Britner & Mossler, 2002; Christiansen & Anderssen, 2010; DePanfilis & Scannapieco, 1994; Noonan & Burke, 2005). At the same time, child problems are ignored in numerous empirical studies into factors influencing out-of-home placement decisions (e.g. Beeman, Kim, & Bullerdick, 2000; Zuravin & DePanfilis, 1997). Contrarily, recent research from the US shows that child characteristics such as emotional and behavioral problems have a stronger impact on placement decisions than do familial and parenting problems. Moreover, comparison of older and more recent data suggests that in recent years emotional and behavioral problems are becoming more important (Bhatt-Sinclair & Sutcliffe, 2012). As a result, the role of child problems and how they relate to a foster placement compared to parenting problems and familial problems remain unclear. In addition, many researchers point out that out-of-home placement decisions are taken under conditions of uncertainty and ambiguity. As a result they are by and large unreliable and rarely rational and/or based on scientific knowledge (Christiansen & Anderssen, 2010). These decisions are strongly influenced by the decision maker’s profession (e.g. juvenile court judges, child protection workers, social workers, and mental health professionals) and attitude toward out-of-home care, availability of out-of-home care services, and (regional) policy regarding placing children in out-of-home care (Arad-Davidzon & Benbenishty, 2008; Britner & Mossler, 2002; Knorth, 1995; Lindsey, 1992). Once the decision for out-of-home care has been made, it must be determined whether the child is to be placed in a foster family or in residential care. The scientific basis for both residential and foster care placement decisions is very weak (Harder, Knorth, & Zandberg, 2006). For both types of decisions, mainly parenting problems such as poor child rearing skills are taken into consideration (Audenaert, 2010; Delfabbro, Barber, & Cooper, 2002). The decision to favor residential care over foster care, is mainly based on the consideration of possible contraindications for foster care. For child welfare workers (and policy makers) foster care is generally the preferred solution, unless there are contraindications such as previous placements (De Meyer, 2002). Strijker and Zandberg (2001) consider behavior by the foster child or his/her parents that can endanger the foster family to be contraindications. Contraindications most often mentioned in Flanders (and the Netherlands) are: delinquency of the foster child or birth parents; excessive use of alcohol or hard drugs; insufficient capabilities for relationship building; extremely deviant behavior, severe physical disabilities, mental disabilities or psychiatric disorders in the child, and opposition to the placement by the birth parents. Finally, Enzlin (1996) states that placement in a foster family is not appropriate for ‘child problems’, when the child’s parents do not accept the foster placement or when the child cannot function within a family. Considering that contraindications are useful but fail to indicate when foster care is appropriate Van Dam, Nordkamp, and Robbroeckx (2000) propose to make foster care placement decisions on the basis of nine central themes: demographic characteristics, placement history, family of origin (relationship with birth parents, acceptance by the parents of the placement. . .), cognitive development, physical development, socio-emotional development, problem behavior, and the need for structure and leisure activities. This ‘variable-oriented approach’ is based on a positive link between these variables and successful foster care placement. The problem with this approach is that it calls for a lot of information that is not always available (Gambrill, 2005). Also, it is likely that not all of this information is needed to arrive at good decisions, because some of these variables are interrelated. Moreover, it is unclear whether each variable is equally important, and how this information should be integrated in order to reach a balanced decision. A ‘profile-oriented approach’ may provide a solution to the abovementioned problems. Such an approach assumes that children with similar characteristics can be grouped into clusters. By clustering foster children with similar characteristics into clearly distinct clinical groups, several variables can be taken into account simultaneously. Also, children can be identified as high-risk at an early stage. Different researchers have attempted to formulate such a profile typology (Delfabbro et al., 2002; Farmer, 1996; Strijker & Zandberg, 1999; Strijker, Zandberg, & Van der Meulen, 2005; Strijker, Zandberg, & Van der Meulen, 2000; Vanderfaeillie, Damen, Pijnenburg, Van den Bergh, & Van Holen, 2014). Strijker and Zandberg (1999) proposed four clusters of foster children using three variables (aggressive behavior, withdrawn behavior, and anxious/depressive behavior). The first (47%) and last cluster (18%) consisted of children with low Please cite this article in press as: Vanderfaeillie, J., et al. Foster care assessment: A study of the placement decision process in Flanders. Child Abuse & Neglect (2015), http://dx.doi.org/10.1016/j.chiabu.2015.04.002

G Model CHIABU-2964; No. of Pages 9

ARTICLE IN PRESS J. Vanderfaeillie et al. / Child Abuse & Neglect xxx (2015) xxx–xxx

3

versus high scores on all three variables respectively. The second cluster (17%) displayed mostly aggressive behavior and low internalizing behavior. Cluster three children (18%) showed the opposite: little aggressive behavior but huge emotional and relational problems (Strijker & Zandberg, 1999). Subsequently these authors proposed a second typology based on behavioral and emotional problems, as assessed by the child behavior checklist (CBCL) (Strijker & Zandberg, 2001; Strijker et al., 2000). Again, four clusters were found: (1) children without any behavioral problems (‘normal’ 39%); (2) children with aggressive and delinquent behavior and social problems (‘aggressive–delinquent’, 24%); (3) children with relatively more internalizing behavior than the second cluster (‘attention and social problems’, 26%), and (4) children with multiple, serious problems in the social sphere (‘withdrawnsocial problems’, 11%). Both typologies proposed by Strijker et al. (2000) are lopsided: they are based solely on behavioral problems of the foster child. This may explain why they were unable to predict a breakdown. Strijker, Zandberg and Van der Meulen (2002) have since developed a third typology, based on the problem behavior of the foster child and parental problems. They identified two clusters. The characteristics of the ‘externalizing’ cluster (31%) are serious problem behavior in the foster child, family conflict and inadequate parenting. The characteristics of the ‘normal’ cluster (69%) are absence of problem behavior in the foster child, moderate family conflict and inadequate parenting due to a lack of parenting skills (Strijker et al., 2002). This typology is significantly associated with breakdown: from the age of four years and two months, children belonging to the externalizing cluster are more at risk for a breakdown. The typology of Strijker et al. (2002) is strikingly similar to that of Farmer (1996) and Delfabbro et al. (2002). On the basis of four variables (foster child’s problem behavior, age and gender, and parental problems), these researchers identified two clusters of foster children: protected and disaffected foster children. Protected foster children are younger and are placed in family foster care because of neglect and abuse. Disaffected foster children are, or remain in family foster care because of serious behavioral and emotional problems, which the birth parents are unable to cope with. Often they are older boys (Delfabbro et al., 2002; Farmer, 1996). Whether this typology was associated with breakdown has not been reported. A typology based on two clusters has recently been replicated in a Flemish–Dutch sample of foster children (Vanderfaeillie et al., 2014). The first protected cluster consisted of younger children for whom parenting and familial problem were more often associated with the placement (77%). The second disaffected cluster consisted of older children where child problems were prominent (23%). Both clusters did not differ in placement history or gender. The study of Vanderfaeillie et al. (2014) has some limitations however. First, the databases used have a different origin, which raises doubts over inter-rater reliability. In addition, data for each database was collected for different purposes. Consequently, inclusion criteria differed between databases. Also, the data was likely not representative for Flanders and the Netherlands. Finally, in contrast to the earlier studies no standardized measure of behavioral problems of the foster children was included. Although previous studies warrant the assumption that the disaffected cluster would have a greater placement history, Delfabbro et al. (2002) and Vanderfaeillie et al. (2014) did not find that the foster child’s placement history differentiated between the two clusters. This hypothesis is first based on the finding that children who are fostered on account of behavioral and emotional problems remain in care longer (Fernandez, 1999). Second, a reciprocal association between problem behavior and breakdown is found, sparking off a negative spiral of breakdowns and behavioral problems (Newton, Litrownik, & Landsverk, 2000; Oosterman, Schuengel, Slot, Bullens, & Doreleijers, 2007). A final argument is the older age of the children in the disaffected cluster. Older children are more at risk of a longer care history, including previous out-of-home placements. In addition, it can be noticed that in terms of size the disaffected clusters in the studies of Strijker et al. (2002) and Vanderfaeillie et al. (2014) differed from those in Delfabbro et al. (2002) and Farmer (1996). Whereas Dutch–Flemish studies found relatively small disaffected clusters (around 30%), in Anglo-Saxon studies this cluster includes almost 50% of the foster children. The studies reviewed above point at the relevance of clustering child and familial characteristics in the context of breakdown prevention and promotion of self-efficacy on the part of foster parents by identifying at-risk children, tailoring foster parent training and support programs and effective foster care placement decision-making. These studies also reveal significant differences between Anglo-Saxon and Flemish-Dutch findings. Considering this and the methodological limitations inherent to the study of Vanderfaeillie et al. (2014), a thorough reiteration and extension of this study is called for. Research questions First, we want to investigate to what extent different problems (child, familial and parenting problems) are associated with placement in foster care. Second, as a ‘profile-oriented’ approach assumes interdependence of the variables used, we question to what extent different problem categories (child, parenting and familial problems) are associated with demographic characteristics of the foster child, including behavioral problems and the child’s placement history. Third, we examine if the finding of two clusters can be replicated in a group of Flemish foster children on the basis of a limited set of demographic foster child characteristics, including the foster child’s placement history and behavioral problems, and problems identified at the time of placement. Method Taking the Vanderfaeillie et al. (2014) study as a reference point, a reiteration (identical data-analytical techniques), expansion (inclusion of a standardized measure of behavioral and emotional problems) and upgrading (use of one Please cite this article in press as: Vanderfaeillie, J., et al. Foster care assessment: A study of the placement decision process in Flanders. Child Abuse & Neglect (2015), http://dx.doi.org/10.1016/j.chiabu.2015.04.002

G Model CHIABU-2964; No. of Pages 9

ARTICLE IN PRESS J. Vanderfaeillie et al. / Child Abuse & Neglect xxx (2015) xxx–xxx

4

representative sample with transparent inclusion criteria and a standardized checklist of problems identified at time of placement) of this latter study has been carried out. Participants and data collection The study sample consists of new family foster care placements with a long-term perspective. Sixteen out of 17 licensed foster care agencies spread across Flanders participated in this study. Within these agencies, all new placements of foster children between 3 and 18 years were surveyed in the fourth month upon placement. Data were collected during one year (October 2010–September 2012). Within this period 504 new foster placements were initiated. Of these 66 (13.1%) were already terminated at the moment of data collection. Of the remaining 438 foster care placements 435 foster mothers (99.3%) and 377 foster care workers (86.1%) filled out the presented questionnaire. For 328 (74.9%) placements both respondents’ questionnaires were filled out; only these were used in the analyses. A comparison of response and non-response group of foster mothers was possible for variables included in the foster care worker’s questionnaire and did not reveal significant differences concerning age, gender and number of previous out-of-home placements of the foster child, type of placement (kinship/non-kinship) and referring authority (voluntary/court ordered) (for all independent sample t-tests and all chi square tests p > .05). Measures Two questionnaires were used. Foster mothers filled out a CBCL measuring behavioral and emotional problems (versions 1.5-6-CBCL/6-18; (Achenbach & Rescorla, 2000, 2001). Since both CBCL versions are only comparable with respect to the broad band scales, only the total problems scale was used. Achenbach and Rescorla (2000) suggest using a T-score >63 as an indicator for problem behavior in the clinical range. The foster care workers filled out a questionnaire designed for the purpose of this study, measuring the following variables: - Characteristics of the foster child: age, gender and number of previous out-of-home placements. The latter variable was transformed into a binary variable (previous out-of-home placement yes/no). - Reasons for placement. With regard to the reasons why children were placed in foster care, foster care workers were asked to select the three most important problems at the time of placement (out of an optional list of 17), such as child abuse, behavioral problems and traumatic life events. These problems were grouped by the researchers into three broad categories: • parenting problems of the birth parent, including poor child rearing, abuse and neglect; • child problems, such as behavioral, socio-emotional and cognitive problems; • familial problems, such as traumatic life events (divorce, serious illness or death of a parent) and problems with housing or finances. Data analysis The three categories of problems identified at the time of the placement, placement history, and gender are binary variables. First descriptive analysis was carried out using frequency distributions and differences were examined (using chi-square tests and logistic regression). Next, logistic regression analyses were done to provide more insight into the ways in which problems associated with the placement were related to of child and placement characteristics. Three regression models with a specific category of problems associated with the placement as the dependent variable were constructed using a two stage block entry. In the first block age, gender and placement history were entered, in the second block problem behavior was added to the model. Finally, a cluster analysis (K-means) was conducted to determine to what extent two clusters of foster children could be found. K-means was adequate because we were a priori looking for two clusters in light of the earlier findings. For the cluster analysis, the three categories of problems identified at the moment of placement, age, gender, problem behavior and placement history were used. Because the results of a cluster analysis can be influenced by differences in scales used, all variables were standardized. Results As presented in Table 1, 135 (41%) of the 328 foster children were in out-of-home care for the first time. 193 (59%) children had already been placed in out-of-home care before, with an average number of previous placements of 2.02 (range 1–8; sd = 1.32). The average age of the children (49% boys; 51% girls) at the start of the placement was nine years (range 2–17; sd = 4.20). Parenting problems (68%) and familial problems (81%) were most often associated with the placement. Child problems (13%) were less frequently cited. Foster children placed for the first time and foster children with a placement history did not differ with respect to the problems associated with placement. However, they differed with respect to gender, age and behavioral problems. Children with a placement history were more likely male, younger and to having more behavioral problems (see Table 1). Please cite this article in press as: Vanderfaeillie, J., et al. Foster care assessment: A study of the placement decision process in Flanders. Child Abuse & Neglect (2015), http://dx.doi.org/10.1016/j.chiabu.2015.04.002

G Model

ARTICLE IN PRESS

CHIABU-2964; No. of Pages 9

J. Vanderfaeillie et al. / Child Abuse & Neglect xxx (2015) xxx–xxx

5

Table 1 Problems at time of placement and characteristics of the foster children. Totaal (N = 328)

First placement (n = 135)

History of out-of-home care (n = 193)

n

%

n

%

n

%

Gender Boy Girl

161 167

49 51

56 79

42 58

105 88

54 46

Parenting problems Familial problems Child problems

223 267 43

68 81 13

92 115 13

68 85 10

131 152 30

68 79 16

Clinical CBCL total problems score Clinical Non-clinical

97 231

30 70

30 105

22 78

67 126

35 65

CBCL total problems Age

2

p

5.308

.021

.003 2.169 2.439 5.953

Mean

sd

Mean

sd

Mean

sd

55.88 9.20

12.63 4.20

53.50 9.79

12.46 4.26

57.54 8.78

12.51 4.12

.96 .141 .118 .015

8.238 4.627

.004 .031

Differences in frequencies between first placements and placements already preceding with out-of-home care were examined with chi-square tests, with exception of differences in age and CBCL total problems score which were analyzed with logistic regression. Table 2 Cross tabulation of child problems and the presence of clinical CBCL total problem score.

Total Problem score

Child problems

Yes

No

Total

Clinical Non-clinical Total

16 27 43

81 204 285

97 231 328

2

1.386

p

.239

Comparison of the presence of behavioral problems as reported by the foster mother and the notion of child problems at the time of placement as reported by the foster care worker showed that ‘child problems’ were mentioned for no more than about half the foster children with behavioral problems. A cross tabulation of ‘child problems’ and the presence a clinical CBCL total problems score was non-significant (2 = 1.39, p = .24). Table 3 shows how the categories of problems identified at the time of placement were associated with the independent variables age, gender and the foster children’s placement history entered in the first block, and with all these variables as well as the foster children’s behavioral problems in the second block. The first model with parenting problems as dependent variable and only the variables age, gender and the foster children’s placement history as independent variables was statistically significant (Nagelkerke R2 = .03, 2 (3) = 7.89, p < .05). However, only age was significantly related to parenting problems (Exp(B) = .99). After addition of the total problems score, the second model was only marginally significant (Nagelkerke R2 = .04, 2 (4) = 8.64, p = .07). Again, in this model only age was related significantly to parenting problems identified at the time of placement (Exp(B) = .99). As in the first model, the odds of parenting problems were higher when the foster child was younger. The odds ratio dropped by 1% ((1.01−1 − 1) × 100%)) with each month the child grew older, or with 12% with each year the child grew older. The first model with child problems as dependent variable and the variables age, gender and the foster children’s placement history as independent variables was statistically significant (Nagelkerke R2 = .10, 2 (3) = 18.34, p < .001). Both variables age (Exp(B) = 1.01) and placement history (Exp(B) = 2.18) were significantly associated with child problems at the time of placement. When the foster child was older at the start of placement or had a placement history, the likelihood of child problems associated with placement was higher. The second child problems-model was statistically significant (Nagelkerke R2 = .13, 2 (4) = 24.02, p < .001) and the addition of the total problems scale added significantly to the model (block 2 (4) = 5.69, p < .05). The notion of child problems was significantly associated with age (Exp(B) = 1.01) and the total problems scale (Exp(B) = 1.04). The odds that child problems coincided with placement increased with 1% each month the child was older and by 4% ((1.04 − 1) × 100%)) with each unit increase (t-scores) of the total problems scale. The other model variables were not significantly related to child problems. The first model with familial problems as dependent variable was only marginally significant (Nagelkerke R2 = .04, 2 (3) = 7.29, p = .07). Table 3 shows that familial problems associated with the placement were only associated with the age of the foster children (Exp(B) = .99). The second familial problems-model proved also marginally significant (Nagelkerke R2 = .04, 2 (4) = 8.38, p = .08); the addition of the total problems scale did not improve the model (block 2 (4) = 1.09, p = .30). The odds of familial problems associated with placement were higher when the child was younger. The likelihood that familial problems coinciding with placement decreased with 1% each month the child was younger. The other model variables, including the Total Problem score, were not related to the familial problems. Table 4 shows that two clusters of foster children could be distinguished using the variables: problems associated with the placement, child age, gender, behavioral problems and placement history. Please cite this article in press as: Vanderfaeillie, J., et al. Foster care assessment: A study of the placement decision process in Flanders. Child Abuse & Neglect (2015), http://dx.doi.org/10.1016/j.chiabu.2015.04.002

G Model

ARTICLE IN PRESS

CHIABU-2964; No. of Pages 9

J. Vanderfaeillie et al. / Child Abuse & Neglect xxx (2015) xxx–xxx

6

Table 3 Predictors of problems associated with the placement (logistic regression, N = 328). Problems coinciding predictor

Parenting problems Exp(B)

p

Wald

Block 1 Age Gender Previous placement Block 2 (df = 3) Model 2 (df = 3)

.99 1.48 .97 7.89 7.89

.020 .106 .897 .048 .048

5.44 2.61 .02

Block 2 Age Gender Placement history Problem behavior Block 2 (df = 1) Model 2 (df = 4)

.99 1.47 .94 1.01 .74 8.64

.022 .111 .795 .388 .388 .071

5.23 2.54 .07 .74

Child problems Exp(B)

Familial problems

p

Wald

Exp(B)

p

Wald

1.01 1.61 2.18 18.34 18.34

.000 .170 .034 .000 .000

12.36 1.88 4.51

.99 .76 .58 7.29 7.29

.047 .338 .073 .063 .063

3.95 .92 3.22

1.01 1.62 1.97 1.04 5.69 24.02

.000 .168 .070 .021 .017 .000

13.71 1.90 3.29 5.33

.99 .76 .60 .99 1.09 8.38

.040 .349 .102 .299 .296 .078

4.23 .88 2.67 1.08

Gender (0 = man, 1 = woman) age (in months).

These two clusters did differ in age (F = 11.03, p < .01), gender (F = 5.41, p < .05), problem behavior of the foster child (F = 4.86, p < .05), child problems (F = 240.52, p < .001) and familial problems (F = 586.88, p < .001). Cluster 1 consisted of older children with more problem behavior. The likelihood of these foster children being girls was also higher. In cluster 1 more often child problems and less often familial problems were identified at the time of placement. Children belonging to cluster 2 could be characterized as younger, with less behavioral problems, and more likely to be boys. Child problems were less often identified at the moment of placement; familial problems on the contrary were identified more frequently. Foster children in both clusters did not differ in terms of placement history or parenting problems associated with placement. Discussion This study explored (a) to what extent different problems (child, familial and parenting problems) are associated with placement in foster care, (b) how foster children’s demographic characteristics, behavioral problems and placement history relate to child, familial and parenting problems, and (c) whether two clusters of foster children could be distinguished in a Flemish sample. Two clusters of foster children were identified. First, a ‘disaffected’ cluster (26%) of older children, more likely girls, whose child problems (reported by foster care workers) and behavioral problems (reported by foster mothers) were prominent. The second, ‘protected’ cluster (74%), consisted of younger children with less behavioral problems, for whom familial problems were more often associated with the foster care placement. These clusters did not differ in placement history or parenting problems associated with the placement. For first time placement foster children and children with a placement history, the main problems associated with the placement were parenting and familial problems. Child problems were cited in only 13% of the cases. Foster mothers reported behavioral problems in 30% of the foster children. Moreover, there was no relation between the mention of child problems at the time of placement and of behavioral problems reported by the foster mothers. Looking at the interplay between child and placement characteristics (age, gender, behavioral problems and placement history) and the problems associated with the placement (parenting, child, and familial problems), it appears that the probability of all these problems being identified at the time of placement was influenced by the foster child’s age. Parenting and familial problems were cited more often for younger foster children; child problems more often for older foster children. Both models with parenting and familial problems as a dependent variable were only marginally significant; the explained variance in both models was around 4%. Problem behavior of the foster children was only associated with a mention of child problems at the time of placement and did not add to the models predicting parenting problems and familial problems. This result indicates that parenting and familial problems exist independent from behavioral problems and cannot be inferred from each other. Consequently, parental problems (parenting and familial problems) and behavioral problems of the foster child should be assessed separately. The results of the cluster analysis (Table 4) largely confirmed the results of the logistic regression analysis (Table 3). The foster children belonging to the first cluster with more child problems at the moment of placement also had more behavioral problems as indicated by a higher total problems score. There was no association however between the mention of child problems at the time of placement and the presence of severe behavioral problems (see Table 2). For this finding several explanations can be put forward. First, it may be the result of poor quality child welfare assessment. Social workers have a tendency to direct their attention more toward the parents than to the children (Christiansen & Anderssen, 2010). Consequently, foster children’s experiences and problems are at risk of going undiagnosed. An assumption on the part of social workers reinforcing this practice is that many see child problems as being the result of parenting and familial problems (Vanderfaeillie et al., 2014). This, in combination with insufficient attention being paid to child problems at the time of a placement (see also above), leads to many foster parents finding themselves confronted with foster children with behavioral Please cite this article in press as: Vanderfaeillie, J., et al. Foster care assessment: A study of the placement decision process in Flanders. Child Abuse & Neglect (2015), http://dx.doi.org/10.1016/j.chiabu.2015.04.002

G Model CHIABU-2964; No. of Pages 9

ARTICLE IN PRESS J. Vanderfaeillie et al. / Child Abuse & Neglect xxx (2015) xxx–xxx

7

Table 4 Results of K-means cluster analysis using the three categories of problems associated with the placement, age, gender, problem behaviour and placement history. Cluster 1 (n = 86) Parenting problems Child problems Familial problems Gender Age (months) Placement history Problem behavior

.14 1.09 −1.34 .21 .30 .10 .20

Cluster 2 (n = 242) −.05 −.39 .48 −.08 −.11 −.04 −.07

F

p

2.216 240.524 586.880 5.409 11.026 1.255 4.860

.138 .000 .000 .021 .001 .263 .028

Gender (0 = man, 1 = woman), placement history (0 = no, 1 = yes).

problems (Armsden, Pecora, Payne, & Szatkiewicz, 2000). They are caught off guard and feel ill-prepared to deal with these behavioral problems. Had they been made aware earlier, they would most likely not have agreed to the placement. Lack of confidence in abilities to handle the behavior of the foster child, as well as lack of support and training are associated with the discontinuation of fostering (Denby, Rindfleisch, & Bean, 1999; Rhodes, Orme, & Buehler, 2001; Whenan, Oxlad, & Lushington, 2009). A second tentative explanation for the high prevalence of foster children with problem behavior is that these problems were not present at time of placement and stem from the trauma caused by the foster care placement (Zeanah et al., 2001). Research shows however that, compared to the time of placement, four months later foster children behave better, are less agitated and less worried, although improvements were small (Barber & Delfabbro, 2005). It was striking that the problems associated with the foster care placement of children with and without a placement history were identical. This suggests that children entered and re-entered out-of-home care due to the same parenting and familial problems. A possible explanation for this observation is that decisions regarding reunification (like placement decisions) are largely unreliable and rarely rational and/or based on scientific knowledge. Moreover, little is known about reunification decision-making and the process of children reintegrating into their family (Wulczyn, 2004). Second, the nonresolving of parenting and familial problems is in line with research showing that for children placed in foster care on account of severe parenting problems such as abuse and neglect or familial problems such as financial or housing problems, the probability of family reunification is smaller and the risk of reentry into foster care after reunification higher (Hayward & DePanfilis, 2007; Miller, Fisher, Fetrow, & Jordan, 2006; Wulczyn, 2004; Wulczyn, Chen, & Orlebeke, 2009). It is important to consider that financial or housing problems, and a lack of professional and informal social support, make these families particularly vulnerable (Fernandez, 2009). Therefore, in order to avoid too long a stay of their children in foster care or reunification failure, intensive support must be offered to birth parents in all domains where they are facing problems. Intensive and concerted services in parenting, mental health, housing, family counseling and substance abuse treatment, and availability of parental legal representation have been shown to improve the likelihood of family reunification and reunification success significantly (Choi & Ryan, 2007; Courtney & Hook, 2012; Lewandowski & Pierce, 2004; Miller et al., 2006). Detailed inspection of both the current and the Vanderfaeillie et al. (2014) cluster solutions tells us that they are very similar. They differ however with respect to two variables: gender and parenting problems. In contrast to the current findings, in the Vanderfaeillie et al. (2014) solution gender did not differentiate between the two groups, whereas parenting problems did. Moreover, the cluster solutions of Farmer (1996) and Delfabbro et al. (2002) also differ from the current solution regarding gender. In both earlier solutions of Farmer (1996) and Delfabbro et al. (2002) girls instead of boys constituted the disaffected group. The question is what relevance should be attached to this finding? Given the fact that in the various related researches gender is volatile factor, differentiating inconsistently, and given the fact that in the logistic regressions gender was not associated with the problems at time of placement, the relevance of gender as differentiating factor is doubtful and warrants further research. In the current study parenting problems do not differentiate between the two clusters. This may not be surprising. In most cases where children are in need of out-of-home care, parenting problems are present at time of placement. Consequently parenting problems will not differentiate between the two groups. Comparison of the current cluster solution with the solutions found in other research with respect to this variable is hindered by the different operationalizations of parenting and familial problems. In contrast to Anglo-Saxon findings (e.g. Delfabbro et al., 2002; Farmer, 1996), yet consistent with earlier Dutch and Flemish research (e.g. Strijker et al., 2002; Vanderfaeillie et al., 2014), the disaffected cluster in the current study is much smaller than the protected cluster. This may be tied to the observation that in Anglo-Saxon countries, foster care is used more often for children with child problems. Research suggests that this is a growing trend (cf. Bhatt-Sinclair & Sutcliffe, 2012). It can be called into question if this is a sound strategy. Using foster care for children with many (behavioral) problems has a negative influence on the effectiveness of family foster care as an intervention, behavioral problems are indeed a strong predictor of a breakdown (Oosterman et al., 2007). This implies that when children are identified as belonging to cluster 1, decision makers should be aware of the increased risk of breakdown when placing these children into family foster care. They should also secure more intensive support for foster parents and/or consider an alternative form of out-of-home care. Please cite this article in press as: Vanderfaeillie, J., et al. Foster care assessment: A study of the placement decision process in Flanders. Child Abuse & Neglect (2015), http://dx.doi.org/10.1016/j.chiabu.2015.04.002

G Model CHIABU-2964; No. of Pages 9

ARTICLE IN PRESS J. Vanderfaeillie et al. / Child Abuse & Neglect xxx (2015) xxx–xxx

8

Our findings stress the importance of approaching foster care assessment and decision-making about additional support as part of a dynamic decision making process. It is important to strive for the most appropriate and feasible situation for the child for the short and longer term (short term foster care followed by a return home, or long-term foster care). At the same time it need to be decided how this situation can be realized (e.g., by providing specific support to the family upon the child’s return home or to the foster parents in order to make the placement sustainable). To arrive at such a dynamic decision process, information is needed about the historical and current situation, in other words: feedback information (e.g.: the foster child’s problem cluster is taken into consideration) as well as feed forward information (e.g. information on the necessary conditions to make a family reunification possible or on the support needed by foster parents to take long-lasting care of the foster child). Feedback information should be gathered using an assessment protocol, in order to identify objectively and in detail the strengths and the weaknesses of the children, the families and their social context. In this process standardized and psychometrically valid observations, (semi-)structured interviews and self-report measures should be used. Based on feedback information it can be decided if a family foster care placement is suitable for this child and his family at this point in time. However, also the purpose of the foster care placement should be considered. This requires the use of feed forward information. Only careful assessment of parenting, familial and child problems, and consequently a thorough understanding of child and family functioning, enable foster care workers to judge which essential advances are needed for successful reunification, how likely these are to occur, and what support services they demand. This evaluation process should be fueled by information on individual characteristics of the foster child and its parents, evidence from outcome research into services promoting reunification and studies on successful reunifications. Dependent on the purpose decided, additional support must focus on strengthening birth parents’ child rearing skills and helping them solve their issues in other life domains and/or in parental role differentiation (Haans, Robbroeckx, Hoogeduin, & Van Beem-Kloppers, 2004): helping parents understand the arguments in favor of long term out-of-home placement and coming to terms with the implications of such a scenario. It is not enough to know which children are well suited for placement in family foster care, and which are not. Conjointly, it must be determined how the birth family can be equipped for successful reunification with their child or – if this not in the child’s best interest – under which conditions a continued (long-term) foster care placement is the best solution for child, foster parents and birth parents. As attitudes toward removal may contribute to higher risk assessments and more intrusive intervention recommendations, decision makers should at the same time be aware of their attitudes and of the danger that intervention decisions may be biased by their personal ideology and values (Davidson-Arad & Benbenishty, 2010). This study has some limitations, the first being, as mentioned above, that the behavioral problems of the foster child were reported by the foster mothers, foster fathers’ opinions are not known. Most likely this is not a serious issue however: in his review Van Holen (2013) found no foster mother–foster father differences in this context. Second, it cannot be excluded that the foster care placement caused the behavioral problems. In light of the limited sensitivity and specificity of the CBCL as screening instrument, the present findings should be interpreted with caution. Third, foster care workers were allowed to report no more than three problems associated with the placement. In cases of multifaceted parenting and familial problems in combination with child problems, this may have led to the registration of mere parental and familial problems, with exclusion of child problems. Finally, the checklist used to assess the problems associated with the placement, was developed by the researchers and awaits validation. In addition, the information used is based on a questionnaire filled out by foster care workers. This might have been different when case workers would have used other standardized sources of information. Conclusion Two clusters of foster children were distinguished. Problems mentioned at time of placement of children with or without a placement history did not differ. This finding suggests that current placement and/or reunification decisions are invalid and that foster care does not succeed in bringing about sufficient improvement of the home environment. It stresses the importance of intensive support of birth parents in all domains where they are facing problems, while also underlining the relevance of dynamic decision making in the assessment of children who are in need of out-of-home care. Implementation of such a dynamic decision process will contribute to better decisions regarding placement in out-of-home care, and subsequently to more permanency and more effective support geared to the different needs of the children and (foster) families involved. References Achenbach, T. M., & Rescorla, L. A. (2000). Manual for ASEBA preschool forms & profiles. Burlington: ASEBA. Achenbach, T. M., & Rescorla, L. A. (2001). Manual for the Aseba school-age forms & profiles: Child behavior checklist for ages 6–18. Burlington: ASEBA. Arad-Davidzon, B., & Benbenishty, R. (2008). The role of worker’s attitudes and parent and child wishes in child protection worker’s assessments and recommendation regarding removal and reunification. Children and Youth Services Review, 30, 107–121. Armsden, G., Pecora, P. J., Payne, V. H., & Szatkiewicz, J. P. (2000). Children placed in long-term foster care: An intake profile using the child behavior checklist/4–18. Journal of Emotional and Behavioral Disorders, 8, 49–64. Audenaert, V. (2010). Pleegzorg: wanneer? Deel 2. Voor welk kind is pleegzorg de beste optie? Een literatuurstudie naar predictoren voor voortijdig afgebroken pleegzorgplaatsingen en indicatiestelling voor pleegzorg. Brussel: Kenniscentrum WVG. Departement Welzijn, Volksgezondheid en Gezin. Barber, J. G., & Delfabbro, P. H. (2005). Children’s adjustment to long-term foster care. Children and Youth Services Review, 27, 329–340.

Please cite this article in press as: Vanderfaeillie, J., et al. Foster care assessment: A study of the placement decision process in Flanders. Child Abuse & Neglect (2015), http://dx.doi.org/10.1016/j.chiabu.2015.04.002

G Model CHIABU-2964; No. of Pages 9

ARTICLE IN PRESS J. Vanderfaeillie et al. / Child Abuse & Neglect xxx (2015) xxx–xxx

9

Barber, J. G., Delfabbro, P. H., & Cooper, L. L. (2001). The predictors of unsuccessful transition to foster care. Journal of Child Psychology and Psychiatry, 42, 785–790. Beeman, S. K., Kim, H., & Bullerdick, S. K. (2000). Factors affecting placement of children in kinship and nonkinship foster care. Children and Youth Services Review, 22, 37–54. Bhatt-Sinclair, K., & Sutcliffe, C. (2012). What determines the out-of-home placement of children in the USA? Children and Youth Services Review, 34, 1749–1755. Britner, P. A., & Mossler, D. G. (2002). Professional’s decision-making about out-of-home placements following instances of child abuse. Child Abuse and Neglect, 26, 317–332. Choi, S., & Ryan, J. P. (2007). Co-occurring problems for substance abusing mothers in child welfare: Matching services to improve family reunification. Children and Youth Services Review, 29, 1395–1410. Christiansen, O., & Anderssen, N. (2010). From concerned to convinced: Reaching decisions about out-of-home care in Norwegian Child Welfare Services. Child and Family Social Work, 15, 31–40. Courtney, M. E., & Hook, J. L. (2012). Evaluation of the impact of enhanced parental legal representation on the timing of permanency outcomes for children in foster care. Children and Youth Services Review, 34, 1337–1343. Davidson-Arad, B., & Benbenishty, R. (2010). Contribution of child protection workers’ attitudes to their risk assessments and intervention recommendations: A study in Israel. Health and Social Care in the Community, 18, 1–9. De Meyer, R. (2002). Indications for foster care in the Netherlands. An evaluation and model proposal. Nijmegen: Katholieke Universiteit, Orthopedagogiek vakgroep Gezin en Gedrag. Delfabbro, P. H., Barber, J. G., & Cooper, L. (2002). Children entering out of home care in South-Australia: Baseline analyses for a 3-year longitudinal study. Children and Youth Services Review, 12, 917–932. Denby, R., Rindfleisch, N., & Bean, G. (1999). Predictors of foster parents’ satisfaction and intent to continue to foster. Child Abuse and Neglect, 23, 287–303. DePanfilis, D., & Scannapieco, M. (1994). Assessing the safety of children at risk of treatment: Decision-making models. Child Welfare, 73, 229–245. Enzlin, P. (1996). Pleegzorg: profiel van een alternatief in de jeugdhulpverlening. Tijdschrift voor Orthopedagogiek Kinderpsychiatrie en Klinische Kinderpsychologie, 21, 13–22. Farmer, E. (1996). Family reunification with high risk children: Lessons from research. Children and Youth Services Review, 18, 403–418. Fernandez, E. (1999). Pathways in substitute care: Representation of placement careers of children using event history analysis. Children and Youth Services Review, 21, 177–216. Fernandez, E. (2009). Children’s wellbeing in care: Evidence from a longitudinal study of outcomes. Children and Youth Services Review, 31, 1092–1100. Gambrill, E. D. (2005). Decision making in child welfare: Errors and their context. Children and Youth Services Review, 27, 347–452. Haans, G., Robbroeckx, L., Hoogeduin, J., & Van Beem-Kloppers, A. (2004). Ouderbegeleiding bij roldifferentiatie. In Ouders helpen bij het invullen van de ouderrol na plaatsing van hun kind in een pleeggezin. Amsterdam: SPW. Harder, A. T., Knorth, E. J., & Zandberg, T. (2006). Residentiele jeugdzorg in beeld. In Een overzichtsstudie naar de doelgroep, werkwijzen en uitkomsten. Amsterdam: SPW. Hayward, R. A., & DePanfilis, D. (2007). Foster children with an incarcerated parent: Predictors of reunification. Children and Youth Services Review, 29, 1320–1334. Hermanns, J. (2010). Pleegkinderen op weg naar volwassenheid. In P. van den Bergh, & T. Weterings (Eds.), Pleegzorg in perspectief (pp. 105–110). Assen: Van Gorcum. Knorth, E. (1995). Besluitvorming over uithuisplaatsing in de jeugdzorg. Kind en Adolescent, 16, 59–63. Lewandowski, C. A., & Pierce, L. (2004). Does family-centered out-of-home care work? Comparison of a family-centered approach and traditional care. Social Work Research, 28, 143–151. Lindsey, D. (1992). Reliability of foster care placement decision: A review. Research on Social Work Practice, 2, 65–80. Miller, K. A., Fisher, P. A., Fetrow, B., & Jordan, K. (2006). Trouble on the journey home: Reunification failers in foster care. Children and Youth Services Review, 28, 260–274. Newton, R. R., Litrownik, A. J., & Landsverk, J. A. (2000). Children and youth in foster care: Disentangling the relationship between problem behaviors and number of placements. Child Abuse and Neglect, 24, 1363–1374. Noonan, K., & Burke, K. (2005). Termination of parental rights: Which foster care children are affected? The Social Science Journal, 42, 241–256. Oosterman, M., Schuengel, C., Slot, W., Bullens, R., & Doreleijers, T. (2007). Disruptions in foster care: A review and meta-analysis. Children and Youth Services Review, 29, 53–76. Rhodes, K. W., Orme, J. G., & Buehler, C. (2001). A comparison of family foster parents who quit, consider quitting, and plan to contiune fostering. The Social Service Review, 75, 84–114. Strijker, J., & Zandberg, T. (1999). Typologie van pleegkinderen. Tijdschrift voor Orthopedagogiek, 38, 385–398. Strijker, J., Zandberg, T., & Van der Meulen, B. (2005). Typologies and outcomes for foster children. Child and Youth Care Forum, 34, 43–55. Strijker, J., & Zandberg, T. J. (2001). Matching in de pleegzorg. In Hulpvraag versus aanbod.. Amsterdam: SPW. Strijker, J., Zandberg, T. J., & Van der Meulen, B. F. (2000). Taxonomie van probleemgedrag van pleegkinderen in pleeggezinnen. Tijdschrift voor Orthopedagogiek, Kinderpsychiatrie en Klinische Kinderpsychologie, 25, 126–135. Strijker, J., Zandberg, T. J., & Van der Meulen, B. F. (2002). Indicators for placement in foster care. British Journal of Social Work, 32, 217–231. Van Dam, W., Nordkamp, S., & Robbroeckx, L. (2000). Passen en meten in pleegzorg. In Plaatsingsmethodiek nader onderzocht. Amsterdam: SWP. Van Holen, F. (2013). Ontwikkeling en implementatie van een trainingsprogramma voor pleegouders gebaseerd op Geweldloos Verzet. Brussel: Vrije Universiteit Brussel (Unpublished doctoral dissertation). Van IJzendoorn, R. (2010). Gehecht aan pleegouders. In P. van den Bergh, & T. Weterings (Eds.), Pleegzorg in perspectief. Ontwikkelingen in theorie en praktijk (pp. 14–23). Assen: Van Gorcum. Vanderfaeillie, J., Damen, H., Pijnenburg, H., Van den Bergh, P., & Van Holen, F. (2014). Foster care assessment: An exploratory study of the placement assessment process in Flanders and the Netherlands. Child and Family Social Work, http://dx.doi.org/10.1111/cfs.12152 Vanderfaeillie, J., & Van Holen, F. (2010). Het verloop van pleeggezinplaatsingen. In P. van den Bergh, & T. Weterings (Eds.), Pleegzorg in perspectief. Ontwikkelingen in theorie en praktijk (pp. 177–193). Assen: Van Gorcum. Vanderfaeillie, J., Van Holen, F., & Coussens, S. (2008). Why do foster care placements break down? A study into the factors influencing foster care placement breakdown in Flanders. International Journal of Child and family Welfare, 11, 77–87. Vanderfaeillie, J., Van Holen, F., Vanschoonlandt, F., Robberechts, M., & Stroobants, T. (2013). Children placed in long-term family foster care: A longitudinal study into the development of problem behavior and associated factors. Children and Youth Services Review, 35, 587–593. Whenan, R., Oxlad, M., & Lushington, K. (2009). Factors associated with foster care well-being, satisfaction and intention to continue providing out-of-home care. Children and Youth Services Review, 31, 752–760. Wulczyn, F. (2004). Family reunification. The Future of Children, 14, 95–113. Wulczyn, F., Chen, L., & Orlebeke, B. (2009). Evaluating contract agency performance in achieving reunification. Children and Youth Services Review, 31, 506–512. Zeanah, C. H., Larrieu, J. A., Heller, S. S., Valliere, J., Hinshaw-Fuselier, S., & Aoki, Y. (2001). Evaluation of a preventive intervention for maltreated infants and toddlers in foster care. Journal of the American Academy of Child and Adolescent Psychiatry, 40, 214–221. Zuravin, S., & DePanfilis, D. (1997). Factors affecting foster care placement of children receiving child protective services. Social Work Research, 21, 34–44.

Please cite this article in press as: Vanderfaeillie, J., et al. Foster care assessment: A study of the placement decision process in Flanders. Child Abuse & Neglect (2015), http://dx.doi.org/10.1016/j.chiabu.2015.04.002