Abstract
Objective This study aimed to identify children’s long-term placement trajectories following early child welfare involvement and the association of these trajectories with subsequent physical and behavioral well-being. Method Participants were 330 children who entered out-of-home care following a substantiated report of child abuse or neglect during infancy/early childhood and their caregivers. Participants were interviewed at child ages 4 and 12 years to assess children’s physical and behavioral well-being and every 2 years in between to determine child placements. Results Latent Class Analyses identified four stable placement trajectories (i.e., adopted [32%], kinship care [15%], stable reunified [27%], and stable foster care [9%]), and two unstable trajectories (i.e., disrupted reunified [12%] and unstable foster care [5%]). Logistic regressions revealed that children in the unstable trajectories had significantly poorer physical and behavioral well-being than children in stable trajectories. Conclusions and Relevance Maltreated children placed in out-of-home care are at risk for long-term placement instability and poorer physical and behavioral well-being.
Keywords: behavior problems, child maltreatment, child trauma, child welfare placement, foster care, physical well-being
During fiscal year 2012, >600,000 children were served in nonrelative, nonadoptive foster care placements (U.S. Department of Health and Human Services, 2013). The majority of children involved in the child welfare system are victims of maltreatment and many have been exposed to other trauma and related prenatal (e.g., infections, substance use, etc.) and environmental (e.g., poverty) risk factors. Many of these children will experience additional stressors, such as disruptions in living arrangements (e.g., multiple caregiver and placement changes), after entering the child welfare system (Webster, Barth, & Needell, 2000; Wulczyn, Kogan, & Harden, 2003). These disruptions in child placements have been linked to negative behavioral (Aarons et al., 2010; Newton, Litrownik, & Landsverk, 2000; Rubin, O'Reilly, Luan, & Localio, 2007) and physical well-being (Takayama, Wolfe, & Coulter, 1998).
Child Welfare Placement Trajectories
The instability of placements in the child welfare system has been well documented (Oosterman, Schuengel, Slot, Bullens, & Doreleijers, 2007). Many of the previous studies of child welfare placement disruptions were designed to provide a “snapshot” of patterns or trajectories in the number and/or timing of caregiver changes that occurred during a discrete period (e.g., 18 months) while children were continuously in the care of the child welfare system (i.e., a “spell"; James, Landsverk, & Slymen, 2004; Rubin et al., 2007; Wulczyn et al., 2003). For example, based on the timing of disruptions over an 18-month period, James and colleagues (2004) used a person-centered approach to identify four trajectories of children with similar patterns of placement disruptions. These trajectories were characterized by: (1) early stability (i.e., in the first 45 days in care), (2) late stability (i.e., after >45 days in care), (3) no discernable pattern of stability, and (4) heterogeneous patterns with some indication of stability. Wulczyn and colleagues (2003) found similar trajectories when they used a latent variable mixture modeling technique in a longer-term (2–4 years) study of placement changes among a cohort of children in foster care in one city. Usher, Randolph, and Gogan (1999) used a more qualitative approach to analyze administrative placement records over a 3- to 4-year period and identified every possible placement trajectory (i.e., >100 trajectories) for all children placed in either kinship or nonrelative foster care in one county during a single year.
In addition to the rich and complex literature documenting children’s placement trajectories while formally involved in the child welfare system, researchers have noted that many children reenter the child welfare system after long-term placements are disrupted and some fail to ever attain stable placements (Kimberlin, Anthony, & Austin, 2009). Researchers have identified predictors of permanent placement disruptions and subsequent reentry in the child welfare system, such as shorter initial stays in out-of-home care, placement in nonrelative foster care rather than kinship foster care, and being discharged to biological parents rather than relatives (Johnson-Reid, 2003; Shaw, 2006; Wulczyn, Hislop, & Harden, 2002). However, no previous studies have identified children’s long-term placement trajectories and none have included children who have exited the child welfare system. These trajectories remain important, as many children who are reunified with their biological parents reenter the child welfare system and others may move frequently among relatives or other informal placements without ever attaining stable living arrangements.
Consequences for Physical and Behavioral Well-Being
Researchers have suggested that disruptions in attachment caused by repeated separations from caregivers and unstable living arrangements increase children’s risk for physical and behavioral problems (Bruskas, 2008). Indeed, children with more child welfare placement disruptions have more physical health (Takayama et al., 1998) and behavioral problems (Aarons et al., 2010; Newton et al., 2000; Rubin et al., 2007), higher mental health service utilization and costs (Rubin et al., 2004), and poorer continuity of health care (DiGiuseppe & Christakis, 2003). However, the impact of long-term placement instability after early child welfare involvement on children’s physical and behavioral well-being has not been previously investigated.
The Present Study
The present study aimed to: (1) use a person-centered latent variable modeling procedure, Latent Class Analysis (LCA), to identify long-term trajectories of placement types and caregiver changes over an 8-year period, among a sample of maltreated children who entered the child welfare system during infancy/early childhood, and (2) examine the physical and behavioral well-being of children in each trajectory following this 8-year period. Differences in the characteristics of children in each placement trajectory were also examined. Based on the previous literature, we expected that (1) both unstable and stable trajectories would be identified among children who were reunified with their biological parents or remained in nonrelative foster care, while only stable trajectories would be identified among children who were adopted or placed in kinship care; (2) children in more unstable trajectories would have poorer physical and behavioral well-being than children in more stable trajectories.
Methods
Sample
The present study included data from the Southwestern site of the Longitudinal Studies of Child Abuse and Neglect (LONGSCAN; Runyan et al., 1998). At this site, a sample of children and caregivers (n = 330) was recruited from the population of children in San Diego County who: (1) had a substantiated report of abuse or neglect during an 18-month enrollment period (May 1990 through October 1991), (2) entered the child welfare system before the age of 3.5 years, and (3) remained in out-of-home care for at least 5 months. By age 4 years, 36% were reunified with their biological parents, 21% were adopted, 19% were living with relatives, and 24% remained in nonrelative out-of-home placements. The sample is diverse (see Table I for sample demographics) and did not differ from the target population on gender, race/ethnicity, or initial placement types based on previous analyses (not presented here).
Table I.
Characteristics of Children in Each Placement Trajectory
| Total sample | Adopted | Kinship care | Stable foster care | Stable reunified | Disrupted reunified | Unstable foster care | ||
|---|---|---|---|---|---|---|---|---|
| Female | n(%) | 174 (53) | 64 (60) | 22 (46) | 15 (50) | 44 (49) | 21 (54) | 8 (47) |
| Race/ethnicity | ||||||||
| White | n(%) | 129 (39) | 50 (47) | 12 (25) | 9 (30) | 43 (48) | 11 (28) | 4 (24) |
| Black | n(%) | 129 (39) | 36 (34) | 26 (54) | 14 (47) | 28 (31) | 17 (44) | 8 (47) |
| Latino/Hispanic | n(%) | 62 (19) | 19 (18) | 10 (21) | 5 (17) | 15 (17) | 8 (21) | 5 (29) |
| Asian/Other | n(%) | 10 (3) | 1 (1) | 0 (0) | 2 (6) | 4 (4) | 3 (7) | 0 (0) |
| Initial removal | ||||||||
| Reason for removal | ||||||||
| Neglect | n(%) | 147 (45) | 63 (60) | 26 (54) | 12 (40) | 27 (30) | 15 (40) | 4 (23) |
| Sexual abuse | n(%) | 22 (7) | 1 (1) | 3 (6) | 2 (7) | 10 (11) | 2 (5) | 4 (23) |
| Physical abuse | n(%) | 56 (17) | 15 (14) | 4 (9) | 5 (17) | 21 (23) | 8 (21) | 3 (18) |
| Other | n(%) | 103 (31) | 26 (25) | 15 (31) | 11 (36) | 32 (36) | 13 (34) | 6 (36) |
| In utero substance exposure | n(%) | 83 (25) | 43 (41) | 11 (23) | 5 (17) | 13 (14) | 8 (21) | 3 (18) |
| Age at first removal | M(SD) | 1.2 (1.2) | .64 (.92) | 1.3 (1.3) | 1.2 (1.2) | 1.6 (1.2) | 1.5 (1.2) | 2.0 (1.2) |
| Placements (initial 18 months) | M(SD) | 3.8 (1.4) | 3.4 (1.2) | 3.8 (1.5) | 3.4 (1.6) | 4.0 (1.4) | 4.1 (1.4) | 4.5 (1.6) |
| Caregiver changes ages 4–12 years | M(SD) | .9 (1.1) | .7 (.6) | 1.0 (1.0) | .4 (.7) | .7 (.8) | 1.9 (1.0) | 3.2 (.6) |
Note. M = trajectory mean; SD = trajectory standard deviation; n = trajectory size. Reasons for removal do not sum to 100% because they were not mutually exclusive.
Procedure
Consistent with LONGSCAN protocol, face-to-face interviews with children and their caregivers were conducted every 2 years beginning when the children were approximately 4 years until they were approximately 12 years old (i.e., between 1992 and 2005) to gain information about the physical, behavioral, and emotional development of the children and the characteristics of their family environments. Caregivers were required to have known the child for a minimum of 6 months before completing each interview. Following San Diego State University institutional review board approval, informed consent was obtained from caregivers, and assent was obtained from the children before conducting each interview. Children and their caregivers were paid a fixed, nominal amount for their participation at each interview.
Administrative child welfare records were used to identify the initial reason for child welfare involvement, the child's age at initial placement, and the number of placements during the first 18 months in care (i.e., before permanent placement). At each interview and continuously through administrative record reviews, children’s placements from ages 4 to 12 years were tracked to determine (a) the type of placement (i.e., adopted, reunified with biological parents, kinship care, nonrelative foster care, or group home) and (b) whether the child was with the same primary caregiver who was identified at the previous interview.
Measures
Child Physical Well-Being
Children (at age 12 years) and caregivers (at ages 4 and 12 years) rated the child’s physical well-being using single global health items, “How is your health,” and “Compared to other children, how is (child’s name)’s health,” respectively, on a 4-point scale (1 = “Excellent,” 2 = “Good,” 3 = “Fair,” 4 = “Poor”). These items were developed for use in the LONGSCAN study and were based on validated self-rated global health items used in previous studies (e.g., Idler & Benyamini, 1997). Data were positively skewed, so a median split was performed to dichotomize each health variable (i.e., 0 = “Excellent” and 1 = “Good to Poor”).
Behavioral Well-Being
Children (at age 12 years) and their caregivers (at ages 4 and 12 years) rated the child’s behavioral well-being using the Youth Self-Report (YSR) and the Child Behavior Checklist (CBCL; Achenbach, 1991), respectively. The CBCL and YSR assess the frequency (0 = not true, 1 = sometimes true, 2 = often true) of 113 child behaviors during the previous 6 months. These items form broadband scales representing child internalizing (e.g., depression, anxiety) and externalizing (e.g., aggression, delinquency) behavior problems. The CBCL and YSR have been empirically validated and internationally normed, have excellent reliability (Cronbach’s αs = .91 and .92 for Internalizing and Externalizing Behavior Problems scales, respectively), and include T-scores for each scale with cutoff scores, which were used in the present study to indicate clinically elevated problems (Achenbach, 1991).
Data Analysis
LCA
LCA in Mplus version 7.12 (Muthén & Muthén, 2012) was used to identify unobserved classes (referred to throughout as trajectories) of children with similar patterns of (a) placement types (i.e., reunified, adopted, kinship or non-relative foster care, or group home) at the age 4, 6, 8, 10, and 12 interviews and (b) caregiver changes between interviews. The goal of LCA is to maximize homogeneity within and heterogeneity between trajectories.
Model Selection and Fit Indices
According to current recommended model selection procedures (Collins & Lanza, 2010; Nylund, Asparouhov, & Muthén, 2007), models with increasing numbers of trajectories were fit sequentially and the Bayesian Information Criterion (BIC; Schwartz, 1978) and sample size-adjusted BIC (Sclove, 1987) were compared until a minimum value of each index was reached. In addition, Collins and Lanza (2010) highlight the importance of considering the theoretical interpretability of each model when selecting the best fitting model. Finally, entropy provided an index of trajectory separation, with values >.80 indicating good and values closer to 1 indicating excellent separation.
Model Parameters
LCA includes two important parameters, Conditional Response Probabilities (CRPs) and Latent Class Probabilities (LCPs). CRPs are estimated for individuals in each trajectory and represent the probability that an individual in trajectory k was in each category (e.g., adopted) of each indicator variable (e.g., age 6 placement type). CRPs can be considered within and between trajectories, as well as relative to overall sample probabilities, to label and substantively differentiate each trajectory from the other trajectories. LCPs represent the probability that each case will be assigned to each trajectory and can be averaged to indicate the relative prevalence of each trajectory in the sample (Collins & Lanza, 2010).
Logistic Regressions
Logistic regression analyses were used to identify differences between placement trajectories in physical and behavioral well-being at age 12 years. Model statistics, odds ratios (ORs), and 95% confidence intervals (CIs) are reported. Dummy-coded variables were created and included in the logistic regression analyses such that each trajectory served as the reference group. This allowed all possible between-trajectory comparisons to be made. Only the significant between-trajectory comparisons are presented here. Age 4 caregiver reports of physical and behavioral well-being were included as covariates in each corresponding model to control for baseline levels of functioning in each domain. Of the 330 participants included in the LCAs, data for 236 (72%) youth and/or caregivers were obtained at age 12 years and were included in these analyses. Analyses (not presented here) revealed no significant differences in baseline demographic characteristics, physical or behavioral well-being, or placement trajectory membership among children with complete and missing data at age 12 years.
Results
Descriptive statistics are reported in Table I. On average, children had 3.8 placement changes (ranging from 0 to 9) during the first 18 months following their initial removal. The present study focused on long-term placement stability (i.e., changes between ages 4 and 12 years in 2-year intervals) and found that children averaged 0.9 (ranging from 0 to 4) caregiver changes.
Placement Trajectories
Descriptive fit criteria for models with one through eight trajectories were compared (see Table II) and suggested that a model with between five and seven trajectories should be considered. Each of these models provided optimal separation of the trajectories (entropy = 0.98). Based on the interpretability, clean structure in the measurement model, and several small trajectories identified by the seven-trajectory model, the six-trajectory model was selected.
Table II.
Latent Class Model Fit Statistics
| Model | BIC | SSA-BIC | Entropy |
|---|---|---|---|
| One trajectory | 5,522.03 | 5,442.73 | N/A |
| Two trajectories | 4,406.93 | 4,245.16 | .99 |
| Three trajectories | 3,964.49 | 3,720.24 | .99 |
| Four trajectories | 3,726.7 | 3,399.99 | .98 |
| Five trajectories | 3,713.22 | 3,304.03 | .98 |
| Six trajectories | 3,741.16 | 3,249.49 | .98 |
| Seven trajectories | 3,819.76 | 3,245.63 | .98 |
| Eight trajectories | 3,909.12 | 3,252.51 | .98 |
Note. BIC = Bayesian Information Criterion; SSA-BIC = sample size-adjusted Bayesian Information Criterion. Bolded text represents the best-fitting model according to each criterion.
CRPs for each placement type at each age and caregiver changes since the previous interview are presented in Table III and descriptive statistics for each trajectory are presented in Table I. Four stable trajectories were identified. Children in the first trajectory, “adopted,” (n = 106 [32%]) had the highest probability of being adopted and remaining with the same caregiver across all ages. However, at age 4 years, they had a combined 0.34 probability of being in either kinship or nonrelative foster care. On average, these children were the youngest at removal (0.64 years or 7.68 months) and had the highest rates of removal for neglect and in utero substance exposure, and lowest rates of removal for sexual abuse. Children in the second trajectory, “kinship care,” (n = 48 [15%]) had the highest probability for being in kinship care and remaining with the same caregiver across all ages. There was some indication of instability across ages, as indicated by probabilities between 0.11 and 0.34 of changing caregivers and probabilities as high as 0.15 for being reunified. This trajectory had the highest proportion of African American children and lowest proportion of children who were removed for physical abuse.
Table III.
Conditional Response Probabilities for Placement Types and Caregiver Stability Across Ages for Each Trajectory
| Sample | Adopted | Kinship care | Stable foster care | Stable reunified | Disrupted reunified | Unstable foster care | |
|---|---|---|---|---|---|---|---|
| N(%) | 330 | 106(32) | 48(15) | 30(9) | 90(27) | 39(12) | 17(5) |
| Age 4: | |||||||
| Reunified | .36 | .01 | .15 | 0 | .79 | .98 | .07 |
| Adopted | .21 | .65 | .02 | 0 | 0 | 0 | 0 |
| Kinship care | .19 | .10 | .79 | .03 | .13 | .03 | 0 |
| Foster care | .24 | .24 | .04 | .97 | .07 | 0 | .87 |
| Group home | 0 | 0 | 0 | 0 | .01 | 0 | .06 |
| Age 6: | |||||||
| Reunified | .38 | 0 | .03 | 0 | .90 | 1 | .17 |
| Adopted | .30 | .94 | .02 | 0 | 0 | 0 | 0 |
| Kinship care | .17 | .03 | .95 | 0 | .07 | 0 | 0 |
| Foster care | .15 | .03 | 0 | 1 | .02 | 0 | .71 |
| Group home | 0 | 0 | 0 | 0 | 0 | 0 | .12 |
| Same caregiver? | .83 | .91 | .66 | .90 | .84 | .92 | .36 |
| Age 8: | |||||||
| Reunified | .35 | 0 | 0 | 0 | 1 | .67 | .13 |
| Adopted | .33 | 1 | 0 | 0 | 0 | .03 | 0 |
| Kinship care | .18 | 0 | .98 | 0 | 0 | .27 | 0 |
| Foster care | .13 | 0 | 0 | 1 | 0 | 0 | .74 |
| Group home | .01 | 0 | .02 | 0 | 0 | .03 | .13 |
| Same caregiver? | .85 | .99 | .86 | .93 | .84 | .67 | .19 |
| Age 10: | |||||||
| Reunified | .32 | 0 | .09 | 0 | 1 | .36 | .07 |
| Adopted | .35 | .99 | .04 | .07 | 0 | 0 | .07 |
| Kinship care | .17 | .01 | .83 | 0 | 0 | .39 | 0 |
| Foster care | .13 | 0 | 0 | .93 | 0 | .20 | .46 |
| Group home | .03 | 0 | .04 | 0 | 0 | .05 | .40 |
| Same caregiver? | .85 | .98 | .83 | 1 | 1 | .32 | .13 |
| Age 12: | |||||||
| Reunified | .29 | 0 | .11 | 0 | 1 | .13 | 0 |
| Adopted | .36 | .98 | .11 | .08 | 0 | .08 | .07 |
| Kinship care | .18 | .02 | .78 | 0 | 0 | .42 | 0 |
| Foster care | .13 | 0 | 0 | .92 | 0 | .20 | .50 |
| Group home | .04 | 0 | 0 | 0 | 0 | .17 | .43 |
| Same caregiver? | .86 | .98 | .89 | 1 | .99 | .43 | .13 |
Note. The highest probabilities for each trajectory at each age are presented in bold.
Children in the third trajectory, “stable reunified,” (n = 90 [27%]) had probabilities of 0.21 and 0.10 for being in out-of-home care at ages 4 and 6 years, but had a high probability of being reunified by age 8 years and thereafter. This trajectory had the highest proportion of children removed for physical abuse and the lowest proportion of children with in utero substance exposure. Children in the fourth trajectory, “stable foster care,” (n = 30 [9%]) had the highest probability of being placed in nonrelative foster care and remaining with the same caregiver across all ages. Although they had small probabilities for changing caregivers at ages 6 and 8 years, they averaged the lowest number of caregiver changes (0.4) between ages 4 and 12 years.
Two unstable trajectories were identified, which included the fifth trajectory of children, “disrupted reunified” (n = 39 [12%]), who initially had the highest probability of remaining reunified, but between ages 8 and 12 years, the majority changed placements. They subsequently had the highest probabilities of being placed in kinship and nonrelative foster care and lowest probabilities of remaining with the same caregiver, and averaged 1.9 placements between ages 4 and 12 years. Finally, children in the sixth trajectory, “unstable foster care,” (n = 17 [5%]) had the highest probability for being in nonrelative foster care and/or group homes and had probabilities between 0.13 and 0.36 for remaining with the same caregivers between each age. They were the oldest at initial removal, had the most initial and long-term placement changes, and included the highest proportion of children who were removed for sexual abuse.
Individuals were assigned to trajectories based on their posterior probabilities for their most likely trajectory membership. Although trajectory memberships are probabilistic, classification accuracy was excellent (i.e., posterior probabilities for most likely trajectories between 0.97 and 1; entropy = 0.98). Moreover, this allowed us to combine smaller trajectories to more evenly balance the trajectory sizes and reduce the number of comparisons that were made in the outcome analyses. On a conceptual basis, the disrupted reunified and unstable foster care trajectories were combined into an “unstable” trajectory and the kinship and stable foster care trajectories were combined into a “stable out-of-home” trajectory.
Physical Well-Being
Approximately 44% of caregivers and 39% of children reported that children’s physical health at age 12 years was worse than “excellent” (i.e., “good” to “poor”). For caregiver reports of children’s physical well-being, the overall model was statistically significant, χ2(4) = 20.05, p < .001, Nagelkerke R2 = 0.12. After controlling for baseline physical well-being, the odds that caregivers reported that children’s physical well-being was less than “excellent” at age 12 years were approximately 2.5 times greater for children in the stable reunified trajectory relative to the adopted trajectory, OR = 2.46, p = .02, 95% CI [1.17, 5.17]. For children’s reports of physical well-being, the overall model was statistically significant, χ2(4) = 11.69, p = .02, Nagelkerke R2 = .07. After controlling for baseline physical well-being, the odds that children reported that their physical well-being was less than “excellent” at age 12 years were more than three times greater for children in the Unstable trajectories relative to the stable reunified trajectory, OR = 3.4, p = .01, 95% CI [1.41, 8.19].
Behavioral Well-Being
Caregivers reported that approximately 17% and 30% of children had clinically elevated Internalizing and Externalizing Problems, respectively, while children reported lower rates (<10%) of both problems. For caregiver reports of children’s Externalizing Problems, the overall model was statistically significant, χ2(4) =20.62, p < .001, Nagelkerke R2 = 0.13. After controlling for baseline Externalizing Problems, the odds that caregivers reported clinically elevated Externalizing Problems at age 12 years were approximately three to five times greater in the adopted, OR = 3.67, p = .01, 95% CI [1.42, 9.44], stable reunified, OR = 3.18, p = .02, 95% CI [1.18, 8.57], and Unstable, OR = 4.71, p = .004, 95% CI [1.63, 13.64], trajectories relative to the stable out-of-home trajectories. For caregiver reports of Internalizing Problems, the overall model was statistically significant, χ2(4) = 14.96, p < .001, Nagelkerke R2 = 0.11. However, after controlling for baseline Internalizing Problems, there were no significant differences in Internalizing Problems at age 12 years between children in different placement trajectories.
For child reports of Externalizing Problems, the overall model was statistically significant, χ2(4) = 9.88, p = .04, Nagelkerke R2 = 0.10. After controlling for baseline Externalizing Problems, the odds of having clinically elevated Externalizing Problems at age 12 years were 5.5 and 6 times greater for children in the Unstable trajectories relative to the stable out-of-home, OR = 5.51, p = .04, 95% CI [1.06, 28.51], and stable reunified, OR = 6.28, p = .03, 95% CI [1.21, 32.58], trajectories, respectively. For child reports of Internalizing Problems, the overall model was statistically significant, χ2(4) = 10.68, Nagelkerke R2 = 0.11. After controlling for baseline Internalizing Problems, the odds of having clinically elevated Internalizing Problems at age 12 years were approximately 6.5 and 7.5 times greater among children in the Unstable trajectories relative to the stable out-of-home, OR = 6.73, p = .02, 95% CI [1.33, 33.98], and stable reunified, OR = 7.47, p = .02, 95% CI [1.48, 37.79], trajectories, respectively.
Discussion
The present study extends prior research by identifying long-term (i.e., 8 years) child placement trajectories among children who entered the child welfare system during infancy or early childhood because of child abuse and/or neglect. Previous studies have considered the number and timing of placement changes that children experienced while formally in the child welfare system (e.g., James et al., 2004; Rubin et al., 2007; Wulczyn et al., 2003). In contrast, the present study was the first to include children who exited the child welfare system (i.e., attained permanent placements) in the identification of six long-term placement trajectories, including: (1) adopted, (2) kinship care, (3) stable foster care, (4) stable reunified, (5) disrupted reunified, and (6) unstable foster care. The present study also adds to prior research by examining differences in physical and behavioral well-being among these trajectories.
Most children had stable placements regardless of placement type. However, an important finding of this study is that a substantial proportion (nearly one-third) of children who were reunified with their biological parents or placed in nonrelative foster care had multiple caregiver changes over their childhoods, which is consistent with the previous literature (e.g., Johnson-Reid, 2003; Rubin et al., 2008; Shaw, 2006; Wulczyn et al., 2002, 2003). Although children in the two trajectories that were reunified with their biological parents were initially removed at similar ages and had similar reasons for removal, the disrupted reunified trajectory was more likely to be reunified initially and then return to out-of-home care, while the stable reunified trajectory was more likely to remain in out-of-home care initially, but remain reunified thereafter. Children in the disrupted reunified trajectory were most likely to be subsequently placed in kinship or nonrelative foster care and had substantial instability in their subsequent placements. These results add to findings from previous short-term studies of reunification failures, which found that shorter stays in out-of-home care are associated with a higher risk for placement disruption (Johnson-Reid, 2003; Shaw, 2006; Wulczyn et al., 2002).
On average, children in the unstable foster care trajectory were the oldest at initial removal, had more placement changes during the initial 18-month period following removal, and had disproportionately high rates of removal for sexual abuse. Each of these factors, as well as early behavior problems, was linked to unstable child welfare placements in a recent meta-analysis (Oosterman et al., 2007). While it is unlikely that the type of maltreatment (e.g., sexual abuse) children experienced would directly affect their long-term placement stability, it is possible that the trauma they endured manifests as acute, severe behavior problems (both internalizing and externalizing; Putnam, 2003; Villodas et al., 2012). When coupled with older age at removal, these factors could place children at risk for chronic placement disruptions and additional adversity. On the other hand, children in the stable foster care trajectory had the fewest average placement changes during the initial 18-month period and between ages 4 and 12 years.
Children in the unstable trajectories generally reported significantly poorer physical and behavioral well-being than children in the stable reunified or stable out-of-home trajectories. This observation extends the findings of short-term studies (Aarons et al., 2010; Newton et al., 2000; Oosterman et al., 2007; Rubin et al., 2007; Takayama et al., 1998), to outcomes of long-term placement instability and suggests a deterioration of physical and behavioral well-being as children encounter additional adversities (Dube, Felitti, Dong, Giles, & Anda, 2003). Differences in placement stability appeared to be more salient to physical and behavioral well-being outcomes than differences in specific placement types. However, it should be noted that no unstable trajectories were identified for children who were initially adopted or placed in kinship care. Although disruptions in these placement types do occur, these placement patterns were not detected in the current sample. While these findings appear to support the conclusion that adoption and kinship care are the ideal long-term placement solutions for maltreated children, more research on the patterns of stability and physical, emotional, and behavioral outcomes of children in these placement types during adolescence is needed. Also, these placement types are not always available to children and additional stable long-term alternative solutions are needed. Although there were some discrepancies between child and caregiver-reported outcomes, this is not uncommon, particularly with regard to somatic and internalizing problems, which may be less easily observed by caregivers than externalizing problems (Cantwell, Lewinsohn, Rohde, & Seeley, 1997; Sourander, Helstelä, & Helenius, 1999). When screening for physical well-being and internalizing problems in particular, ratings from youth should be considered most strongly.
Limitations
The implications of this study should be considered in light of its limitations. Rates of initial child welfare entry are estimated to be at least twice as high among children who enter the child welfare system very early, relative to children ≥4 years (Wulczyn et al., 2002). These results may not directly generalize to children who first enter the child welfare system when they are older. The indicators of placement change used in the present study do not represent all placement changes that children experienced across childhood, but rather whether children experienced changes over each 2-year period, consistent with the focus of the present study on children’s long-term placement trajectories across childhood. Although previous research has supported the reliability and validity of single-item global self-report measures of physical health (Idler & Benyamini, 1997; Lundberg & Manderbacka, 1996), the limited response distributions for these variables in the present study should be considered when interpreting these results and more comprehensive measures should be used in future studies. As in all longitudinal studies, intervening policy changes (e.g., mandates for permanent placement planning) can exert an influence on study outcomes and it is important to consider these findings in the context of changing legislation at the state and federal levels. Finally, although caregivers were required to have known the child for at least 6 months before providing ratings and the validity of group care workers ratings of behavior problems on the CBCL has been established (Albrecht, Veerman, Damen, & Kroes, 2001), it is possible that changing caregivers affected the ratings of physical and behavioral well-being among children in the unstable trajectories.
Implications
The consequences of early child maltreatment may be exacerbated by children’s experiences of instability in out-of-home and long-term placements following early involvement in the child welfare system. Data from this prospective study indicated that a substantial proportion of children who were involved in the child welfare system at an early age because of abuse and/or neglect had chronically unstable placements across childhood, which contributed to poor long-term physical and behavioral well-being. It is important for health-care providers and other professionals working with children who are currently or were formerly involved in the child welfare system to consider: (1) children’s long-term experiences of placement instability, (2) the accumulation of adversity associated with these experiences and its impact on children’s physical and behavioral well-being, and (3) health information from multiple sources, including youth, particularly for physical health and internalizing problems that are not easily observable. The promotion of long-term placement stability and consistent access to physical and mental health services for abused and neglected children should be a primary focus of prevention and early intervention programs. Administrative health and child welfare policies should facilitate the placement stability and continuity of care among children beyond their formal exits from care.
Funding
This work was supported by the Office of Child Abuse and Neglect (Grant numbers 90CA1744, 90CA1745, 90CA1746, 90CA1747, 90CA1748, and 90CA1749).
Conflicts of interest: None declared.
References
- Aarons G. A., James S., Monn A. R., Raghavan R., Wells R. S., Leslie L. K. (2010). Behavior problems and placement change in a national child welfare sample: A prospective study. Journal of the American Academy of Child and Adolescent Psychiatry , 49(1), 70–80. doi: 10.1097/00004583-201001000-00011 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Achenbach T. M. (1991). Integrative manual for the 1991 CBCL/4-18, YSR, and TRF profiles. Burlington, VT: Department of Psychiatry, University of Vermont. [Google Scholar]
- Albrecht G., Veerman J. W., Damen H., Kroes G. (2001). The Child Behavior Checklist for group care workers: A study regarding the factor structure. Journal of Abnormal Child Psychology , 29, 83–89. doi: 10.1023/A:1005255614039 [DOI] [PubMed] [Google Scholar]
- Bruskas D. (2008). Children in foster care: A vulnerable population at risk. Journal of Child and Adolescent Psychiatric Nursing , 21, 70–77. doi: 10.1111/j.1744-6171.2008.00134.x [DOI] [PubMed] [Google Scholar]
- Cantwell D. P., Lewinsohn P. M., Rohde P., Seeley J. R. (1997). Correspondence between adolescent report and parent report of psychiatric diagnostic data. Journal of the American Academy of Child and Adolescent Psychiatry , 36, 610–619. doi: 10.1097/00004583-199705000-00011 [DOI] [PubMed] [Google Scholar]
- Collins L. M., Lanza S. T. (2010). Latent class and latent transition analysis with applications in the social, behavioral, and health sciences. Hoboken, NJ: John Wiley and Sons, inc. [Google Scholar]
- DiGiuseppe D. L., Christakis D. A. (2003). Continuity of care for children in foster care. Pediatrics , 111, e208–w213. [DOI] [PubMed] [Google Scholar]
- Dube S. R., Felitti V. J., Dong M., Giles W. H., Anda R. F. (2003). The impact of adverse childhood experiences on health problems: Evidence from four birth cohorts dating back to 1900. Preventive Medicine , 37, 268–277. [DOI] [PubMed] [Google Scholar]
- Idler E. L., Benyamini Y. (1997). Self-rated health and mortality: A review of twenty-seven community studies. Journal of Health and Social Behavior , 38, 21–37. [PubMed] [Google Scholar]
- James S., Landsverk J., Slymen D. J. (2004). Placement movement in out-of-home care: Patterns and predictors. Children and Youth Services Review , 26, 185–206. [Google Scholar]
- Johnson-Reid M. (2003). Foster care and future risk of maltreatment. Children and Youth Services Review , 25, 271–294. doi: 10.1016/S0190-7409(03)00012-4 [Google Scholar]
- Kimberlin S. E., Anthony E. K., Austin M. J. (2009). Re-entering foster care: Trends, evidence, and implications. Children and Youth Services Review , 31, 471–481. [Google Scholar]
- Lundberg O., Manderbacka K. (1996). Assessing reliability of a measure of self-rated health. Scandinavian Journal of Public Health , 24, 218–224. [DOI] [PubMed] [Google Scholar]
- Muthén L. K., Muthén B. O. (2012). Mplus user’s guide (7th edn). Los Angeles: Muthén & Muthén. [Google Scholar]
- 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. [DOI] [PubMed] [Google Scholar]
- Nylund K. L., Asparouhov T., Muthén B. O. (2007). Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carlo simulation study. Structural Equation Modeling , 14, 569–569. doi: 10.1080/10705510701575396 [Google Scholar]
- Oosterman M., Schuengel C., Slot N. W., Bullens R. A. R., Doreleijers T. A. H. (2007). Disruptions in foster care: A review and meta-analysis. Children and Youth Services Review , 29, 53–76. doi: 10.1016/j.childyouth.2006.07.003 [Google Scholar]
- Putnam F. W. (2003). Ten-year research update review: Child sexual abuse. Journal of the American Academy of Child and Adolescent Psychiatry , 42, 269–278. doi: 10.1097/00004583-200303000-00006 [DOI] [PubMed] [Google Scholar]
- Rubin D. M., Alessandrini E. A., Feudtner C., Mandell D. S., Localio A. R., Hadley T. (2004). Placement stability and mental health costs for children in foster care. Pediatrics , 113, 1336–1341. [DOI] [PubMed] [Google Scholar]
- Rubin D. M., Downes K. J., O'Reilly A. L., Mekonnen R., Luan X., Localio R. (2008). Impact of kinship care on behavioral well-being for children in out-of-home care. Archives of Pediatrics and Adolescent Medicine , 162, 550–556. doi: 10.1001/archpedi.162.6.550 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Rubin D. M., O'Reilly A. L. R., Luan X., Localio A. R. (2007). The impact of placement stability on behavioral well-being for children in foster care. Pediatrics , 119, 336–344. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Runyan D. K., Curtis P. A., Hunter W. M., Black M. M., Kotch J. B., Bangdiwala S., Dubowitz D., English D., Everson M. D., Landsverk J. (1998). LONGSCAN: A consortium for longitudinal studies of maltreatment and the life course of children. Aggression and Violent Behavior , 3, 275–285. doi: 10.1016/S1359-1789(96)00027-4 [Google Scholar]
- Schwartz G. (1978). Estimating the dimensions of a model. The Annals of Statistics , 6, 461–464. doi: 10.1214/aos/1176344136 [Google Scholar]
- Sclove S. L. (1987). Application of model-selection criteria to some problems in multivariate analysis. Pyschometrika , 52, 333–343. doi: 10.1007/BF02294360 [Google Scholar]
- Shaw T. V. (2006). Reentry into the foster care system after reunification. Children and Youth Services Review , 28, 1375–1390. doi: 10.1016/j.childyouth.2006.02.006 [Google Scholar]
- Sourander A., Helstelä L., Helenius H. (1999). Parent-adolescent agreement on emotional and behavioral problems. Social Psychiatry and Psychiatric Epidemiology , 34, 657–663. doi: 10.1007/s001270050189 [DOI] [PubMed] [Google Scholar]
- Takayama J. I., Wolfe E., Coulter K. P. (1998). Relationship between reason for placement and medical findings among children in foster care. Pediatrics , 101, 201–207. [DOI] [PubMed] [Google Scholar]
- U.S. Department of Health and Human Services, Administration for Children and Families, Administration on Children, Youth and Families, & Children's Bureau. (2013). Trend in Foster Care and Adoption (FFY 2002-FFY 2012). [Google Scholar]
- Usher C. L., Randolph K. A., Gogan H. C. (1999). Placement patterns in foster care. Social Service Review , 73, 22–29. [Google Scholar]
- Villodas M. T., Litrownik A. J., Thompson R., Roesch S. C., English D. J., Dubowitz H., Kotch J. B., Runyan D. K. (2012). Changes in youth's experiences of child maltreatment across developmental periods in the LONGSCAN consortium. Psychology of Violence , 2, 325–338. doi: 10.1037/a0029829 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Webster D., Barth R. P., Needell B. (2000). Placement stability for children in out-of-home care: A longitudinal analysis. Child Welfare , 79, 614–632. [PubMed] [Google Scholar]
- Wulczyn F., Hislop K. B., Harden B. J. (2002). The placement of infants in foster care. Infant Mental Health Journal , 23, 454–475. doi: 10.1002/imhj.10028 [Google Scholar]
- Wulczyn F., Kogan J., Harden B. J. (2003). Placement stability and movement trajectories. Social Service Review , 77, 212–236. doi: 10.1086/373906 [Google Scholar]