Child Abuse & Neglect 38 (2014) 498–509

Contents lists available at ScienceDirect

Child Abuse & Neglect

Caregiver instability and early life changes among infantsreported to the child welfare system�

Cecilia Casanuevaa,∗, Mary Dozierb, Stephen Tuellerc, Melissa Doland,Keith Smithd, Mary Bruce Webbe, T’Pring Westbrookf, Brenda Jones Hardeng

a RTI International, 3040 Cornwallis Road, P.O. Box 12194, Research Triangle Park, NC 27709-2194, USAb Amy E. du Pont Chair of Child Development, Department of Psychology, University of Delaware Newark, 114 Wolf Hall, Newark, DE19716, USAc RTI International, 1080 Grandview Drive, Providence, UT 84332, USAd RTI International, 230 W. Monroe, Suite 2100, Chicago, IL 60606-4901, USAe Division of Child and Family Development Office of Planning, Research and Evaluation Administration for Children and Families, U.S.Department of Health and Human Services, 370 L‘Enfant Promenade SW, Washington, DC 20447, USAf Office of Planning, Research and Evaluation Administration for Children and Families, 370 L‘Enfant Plaza Promenade, SW, 7th FloorWest, Washington, DC 20447, USAg Institute for Child Study, Department of Human Development, University of Maryland College Park, 3301 Benjamin Building, CollegePark, MD 20742, USA

a r t i c l e i n f o

Article history:Received 9 May 2013Received in revised form 26 July 2013Accepted 29 July 2013Available online 29 August 2013

Keywords:Caregiver instabilityEarly life changes: InfantsChild welfare systemNSCAW

a b s t r a c t

This study describes the extent of caregiver instability (defined as a new placement for 1week or longer in a different household and/or with a new caregiver) in a nationally rep-resentative sample of infants, followed for 5–7 years. Data were drawn from the NationalSurvey of Child and Adolescent Well-Being (NSCAW), a longitudinal study of 5,501 childreninvestigated for child maltreatment. The analysis sample was restricted to 1,196 infants.Overall, 85.6% of children who were infants at the time of the index maltreatment expe-rienced at least one caregiver instability event during their first 2 years of life. Caregiverinstability was associated with the child having a chronic health condition and the care-giver being older than 40 years of age at baseline. The levels of instability reported in thisstudy from infancy to school entry are extremely high. Children with more risk factors weresignificantly more likely to experience caregiver instability than children with fewer riskfactors. The repeated loss of a young child’s primary caregiver or unavailable, neglectfulcare can be experienced as traumatic. Some evidence-based programs that are designedto work with young maltreated children can make a substantial positive difference in thelives of vulnerable infants.

© 2013 Elsevier Ltd. All rights reserved.

Introduction

The youngest children are the most vulnerable to child maltreatment (U.S. Department of Health and Human Services[DHHS], 2012b) and to the effects of caregiver instability, given their need for consistent and sensitive caregiving to developoptimally (DHHS, 2012a). A paradox of the child welfare system (CWS) is that although some children must be placed out

� The National Survey of Child and Adolescent Well-Being was developed under contract to RTI International from the Administration of Children andFamilies of the U.S. Department of Health and Human Services. Conclusions do not necessarily represent those of the Administration of Children andFamilies.

∗ Corresponding author.

0145-2134/$ – see front matter © 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.chiabu.2013.07.016

oeW

difi

Ws4t3r

caia

obct

agciitria(tfic

P

IdhA2Cicogoa

Ctog2O

C. Casanueva et al. / Child Abuse & Neglect 38 (2014) 498–509 499

f home to avoid further harm, interventions that result in changes of caregivers may disrupt children’s attachments—thenduring emotional bond between an infant and a caregiver that leads to the child’s reliance on an adult (Ainsworth, Blehar,aters, & Wall, 1978)—and generate trauma because of the loss of primary relationships.This study aimed to describe the extent of caregiver instability—defined as a new placement for one week or longer in a

ifferent household and/or with a new caregiver—by combining information from caseworkers and caregivers about infantsnvolved in maltreatment investigations. A better understanding of this phenomenon will help the CWS identify risk factorsor caregiver instability and address these risk factors in a case plan that can enhance caregiver stability and maintain thenfant’s attachments.

To examine caregiver instability, we used five waves of longitudinal data from the National Survey of Child and Adolescentell-Being (NSCAW), a longitudinal study intended to answer a range of fundamental questions about the functioning,

ervice needs, and service use of children who come in contact with the CWS. Although some children in the study (almost0% of infants studied) were placed out of home by child welfare authorities across the five to seven years, more than half ofhe children in this study remained in their homes of origin—and of those left at home at the time of the index maltreatment,5.2% continue to receive CWS services. Consequently, for the purposes of this article, caregiver instability does not necessarilyefer to formal placement in foster care settings.

We investigated multiple aspects of instability, including timing of caregiver instability experiences and cumulativeounts of changes in caregivers from infancy to school entrance, both within and outside the auspices of the CWS. Welso examined possible correlates of instability, including child and family factors, and characteristics of the maltreatmentnvestigation. Additionally, we used two indices that count risk factors at the time of the index maltreatment investigationnd we describe their association with instability.

The study of caregiver instability among maltreated infants is critical and encompasses several complex issues: (a) becausef the loss of the attachment figure, providing a new caregiver can be a source of stress to the infant; (b) a new caregiver cane a healing experience if the new caregiver is able to nurture, protect, and help stabilize the distressed infant; or (c) a newaregiver can be a new source of stress if the caregiver is insensitive to the signs of infant trauma, unable to help regulatehe distress, and does not facilitate the development of attachment (Zeanah, Berlin, et al., 2011).

A secure attachment relationship with a caregiver aids in a child’s development of emotion regulation and self-confidence,llowing the child to learn to function autonomously and competently (Ainsworth et al., 1978). In contrast, insensitive care-iving (e.g., intrusive, rejecting, frightening behaviors) is associated with insecure attachment relationships, which can beharacterized as avoidant, resistant, or disorganized. Each of these insecure patterns of attachment has been linked withnternalizing and externalizing problems, with disorganized attachment being the strongest predictor of later externaliz-ng problems (Fearon, Bakermans-Kranenburg, et al., 2010). Caregiver instability threatens the child’s developing abilityo maintain trust in the attachment relationships, undermining the developmental expectation that the caregiver will beeliably available as a protection from danger. These expectations may result in a host of negative developmental outcomes,ncluding hypervigilance; difficulties in concentrating; recurring play representing the traumatic situation and nightmares;nd constriction of the child’s motivation to play, explore, and learn from the physical and interpersonal environmentLieberman, Chu, et al., 2011). Whereas many studies have made the connection between maltreatment, risk factors,oxic stress, and developmental problems, less is known about the pervasiveness of the experience of losing attachmentgures among infants in the CWS or the association between risk factors and caregiver instability for these vulnerablehildren.

lacement instability in the child welfare system

nfants and toddlers. Infants and toddlers are overrepresented in the CWS. For example, the 2011 annual report based onata from the National Child Abuse and Neglect Data System (NCANDS) stated that infants aged birth to 1 year old had theighest rate of victimization at 21.2 per 1,000 children of the same age group in the national population (DHHS, 2012c).mong children entering foster care in 2011, 48% were aged birth to 5 years old, with 16% younger than 1 year old (DHHS,012a). Very few studies have focused on placement instability (defined as multiple changes in placement known by theWS) among infants. These studies have reported that compared with older children, infants are at increased risk for entry

nto foster care, particularly if they are 0–3 months old (Wulczyn, Hislop, & Harden, 2002). In part, the higher risk of fosterare placement has been associated with in utero exposure to substances (Goerge & Harden, 1993). To our knowledge, onlyne previous study, based on data from 11 states between 1990 and 1997, has analyzed the CWS experiences of specific ageroups within the population of infants. The study (Wulczyn et al., 2002) found higher risk of reentry to foster care if the agef admission at the first foster placement was over 3 months old, the child was African American (compared with white),nd the first placement was in foster care (compared with kin care).

hildren’s characteristics. Previous studies of placement instability have focused mostly on larger age ranges than included inhe present study. Among studies of CWS populations that include infants, risk factors for placement instability include type

f out-of-home placement (e.g., kin care has been associated with lower numbers of placements compared with foster androup homes [Webster, Barth, & Needell, 2000]), gender (males experienced more placements than females; Webster et al.,000), and race (some studies report that white children are more likely to have more placements; Webster et al., 2000).ther studies report disproportionately high rates of placements for African American children (Wulczyn et al., 2002), by

500 C. Casanueva et al. / Child Abuse & Neglect 38 (2014) 498–509

type of maltreatment (e.g., neglect has been associated with a lower number of placements than other types of maltreatment;Webster et al., 2000), health problems (i.e., the child’s health problems are associated with a higher number of placements;Eggertsen, 2008), maltreatment substantiation (i.e., substantiated is associated with more placements; Eggertsen, 2008), andmental health problems in the family of origin (i.e., mental health problems are associated with more placements; Rubin,O’Reilly, Luan, & Localio, 2007).

Children over 2 years old. Among studies of non-infants (i.e., 2 years old or older), child-level predictors of placement insta-bility included older age (Foster, Hillemeier, & Bai, 2011; Rubin et al., 2007; Wulczyn, Kogan, & Harden, 2003), AfricanAmerican race (Foster et al., 2011), emotional and behavioral problems (Foster et al., 2011), sexual abuse (Eggertsen, 2008),and placement with non-kin compared with kin (Foster et al., 2011). Several other studies reported simply on predic-tors of one or more placements (yes/no). These studies found that among the caregiver characteristics associated withplacement are older age (Berger, Bruch, Johnson, James, & Rubin, 2009); low educational attainment (Berger et al., 2009);mental health, substance abuse, or active domestic violence problems (Bhatti-Sinclair & Sutcliffe, 2012); poverty (Bergeret al., 2009; Horwitz, Hurlburt, Cohen, Zhang, & Landsverk, 2011); and higher risk scores among family of origin (Bergeret al., 2009; Horwitz et al., 2011). Among the child protective services (CPS) investigation characteristics associated withplacements are substantiated CPS investigation (Berger et al., 2009) and neglect as main type of maltreatment (Simmel,Morton, & Cucinotta, 2012). Among child characteristics associated with placement are problems with daily living skills(Horwitz et al., 2011), disability (Lightfoot et al., 2011), health problems (Bhatti-Sinclair & Sutcliffe, 2012), being male(Simmel et al., 2012), and being African American (Bhatti-Sinclair & Sutcliffe, 2012; Simmel et al., 2012; Wulczyn et al.,2002). Taken together, these studies demonstrate that in the study of caregiver instability among infants it is important toconsider environmental factors represented by the caregivers’ characteristics, experience with the CWS, and the infant’sfactors.

Risk indices as predictors of child welfare outcomes

To help understand what predicts children’s risk of caregiver instability, interest has been generated in the field for the cre-ation of a risk index that can be used as a predictor of caregiver instability. Of the three main child welfare outcomes—safety,permanency, and well-being—few studies have used a risk index to predict placement and permanency. Two studies usingNSCAW data have used a risk index to predict any removal by the CWS and at least one out-of-home placement across3 years since the index maltreatment. Both studies created a family risk factors index based on 24 items, with informa-tion provided by caseworkers on caregivers’ use of drugs, mental health, parenting, domestic violence, and other riskindicators with each factor coded dichotomously. Children who experienced out-of-home placement had higher familyrisk index scores (Berger et al., 2009; Horwitz et al., 2011). In contrast to the few studies using a risk index to predictplacement, several studies predicting child well-being have used a risk index, but most include only three to five indica-tors of risk. The literature on children’s well-being has linked negative outcomes to some specific risk factors but also tocumulative risk (i.e., the number of risk factors rather than any specific combination of risk factors; Lewis, Dozier et al.,2007).

One of the aims of the present study is to extend these analyses by creating an indicator of caregiver instability thatintegrates information from both caseworkers and caregivers surveyed on NSCAW and using risk indices to predict caregiverinstability among infants. An index that could help the CWS identify children at higher risk of caregiver instability would bevery beneficial to the field. Because the great majority of children remain in their homes after a maltreatment investigation,it is critical for the CWS to know what to expect about children’s stability of care beyond the time of contact with the CWSand to understand which children may need additional support (even if they are never placed in foster care) to promotecaregiver stability.

In this study, we used two risk indices, the NSCAW Index and the Adverse Childhood Experience Index. Whereas theNSCAW Index can provide a quantitative indicator of accumulated risk among maltreated children, it does not include theexperience of maltreatment per se. The index created for the Adverse Childhood Experiences (ACEs) Study (Anda, Felitti,et al., 2006) was also reviewed and used in this study because it includes maltreatment events. The ACEs study is an ongoingcollaboration between Kaiser Permanente and the U.S. Centers for Disease Control and Prevention (CDC). It consists of morethan 17,000 adults aged 18 or older who are members of Kaiser Permanente and were interviewed between 1995 and 1997.The number of adverse childhood experiences (i.e., ACE score) is a measure of cumulative childhood stress that is stronglyassociated among adults with higher risk of nearly every outcome in the affective, somatic, substance abuse, memory, andaggression domains (Anda et al., 2006).

Consequently, using a unique national probability sample of infants referred to the CWS, the analyses for the presentstudy were designed to (a) document the frequency of caregiver instability over a 5- to 7-year follow-up period, (b) identifychild, family, and environmental risk factors related to caregiver instability, and (c) model predictors of caregiver instability

across time. We hypothesized that caregiver instability would be higher among children with more risk factors on the ACEsand NSCAW indices than among children with fewer risk factors. We also expected that individual risk factors that maymake caregiving challenging (e.g., caregiver mental health problems) would be more predictive of caregiver instability thancontextual factors (e.g., receipt of income support).

C. Casanueva et al. / Child Abuse & Neglect 38 (2014) 498–509 501

Table 1Sociodemographic characteristics of infants in the child welfare system.

Child characteristic N Total % SE

Gender 1196Male 49.1 3.2

Age 11960–3 months 24.2 2.24–12 months 75.8 2.2

Race/ethnicity 1190White 43.7 3.8Black 29.6 3.0Hispanic 21.3 2.3Other 5.4 1.2

Setting baseline 1185In home 77.1 2.5Informal kin care 5.0 0.9Formal kin care 3.9 1.0Foster care 8.8 1.1

N

M

Csc(w

tt

P

nfcAyal(

P

hsc

M

dh

Saa2

Other out-of-home arrangement 5.2 1.2

ote. All percentages are weighted, Ns are unweighted.

ethods

NSCAW is a longitudinal study of the well-being of 5,501 children aged 14 years or younger who had contact with theWS during a 15-month period starting October 1999. The sample design involved a stratified two-stage sample, the primaryampling units being county child welfare agencies. Families included both open cases in the CWS and closed cases (i.e., caseslosed without offering of services). The sample design required oversampling of open cases, infants, and sexual abuse casesBiemer, Dowd, & Webb, 2010). Children were included in the sample regardless of whether the index report of maltreatmentas substantiated.

NSCAW data were collected from children, their caregivers, and caseworkers at baseline (approximately 4 months afterhe index investigation) and follow-up points conducted at 1 year, 1.5 years, 3 years, and 5–7 years after the completion ofhe maltreatment investigation. Additional information on NSCAW methods appears elsewhere (Biemer et al., 2010).

articipants

The analysis for this study focuses on 1,196 children who were infants at the time of the index investigation for abuse oreglect and were followed up until they were 5–7 years old. The data used here were collected from 1999 to 2007 and drawn

rom interviews of caregivers and caseworkers for all children who were 12 months or younger at baseline in the NSCAWhild protective services sample. Response rates for the subsample of infants included in this study were 69.7% at baseline.mong those who responded at baseline, 88.3% responded at 1 year, 88.8% at 1.5 years, 85.4% at 3 years, and 84.4% at 5–7ears. Table 1 provides general information about the infants. Almost half of these infants investigated for maltreatment hadcase that was substantiated (38.0%) or indicated (8.3%), meaning the CWS decided that the allegations of maltreatment had

egal evidence (substantiated) or that some finding of maltreatment existed (indicated) but not enough for substantiationDrake, 1996).

rocedures

Field representatives were selected from candidates with experience with the CWS population and with the ability toandle stressful situations. Field representatives contacted caregivers and asked permission to interview them about theelected child and to assess the child directly by means of standardized measures. Baseline interviews and assessments wereonducted an average of 4 months after the CWS investigation for maltreatment.

easures

NSCAW was advised by a panel of national experts to determine the list of questions, instruments, and tests used at eachata collection wave. Whenever possible, standardized instruments with national samples, or instruments or questions thatad been used in previous studies with large and diverse national samples of children and families, were chosen.

ociodemographics. Caregivers were asked about their child’s sex, age, race/ethnicity, family income, and number of adultsnd children in the household, as well as the caregivers’ age, education, and marital status. Family income and number ofdults and children in the household were used to determine poverty level, based on U.S. Census Bureau guidelines (Dalaker,001).

502 C. Casanueva et al. / Child Abuse & Neglect 38 (2014) 498–509

Fig. 1. Timeline for one infant on the trajectory of placement since the time of the index report to the Child Welfare System, as reported by caseworkers andcaregivers, in months. Note. Caseworker = Caseworker report of change in placement; Caregiver = Caregiver report of a change of placement. To be counted,instability was defined as changes in caregivers and/or households lasting 1 week or longer.

Instability of living arrangements. At each data collection wave, information on the child’s living arrangements was obtainedfrom caseworker and caregiver interviews. Questions about changes in placement were repeated for up to 25 moves sincethe last interview for both sources.

Instability was defined as changes at any point after the baseline interview in the child’s household (described by case-workers responding to “Where is the child currently living?”) or caregiver (described by caregivers responding to “Has theliving situation changed?” that in a small number of cases included the option of “parent/guardian returned home,” whichmay mean a change in caregiver but not a change in household). When caseworkers’ and caregivers’ reports of placementsdiffered by less than 1 month and the caregiver type did not change, reflecting the same placement, information providedby the caseworkers was given priority. Following a previous study (Howard, Martin, Berlin, & Brooks-Gunn, 2011), a changein a child’s caregiver was counted if the child was in a new household for 7 days or more and the original caregiver didnot move to the new household with the child. We counted the child’s caregiver and household at the time of the baselineinterview as “0”; that is, if a child was already in foster care at baseline, that placement was not included in the count.

During the process of creating derived variables to represent caregiver instability, we created a timeline plot for eachchild. Plots graphically represented information coming from the caseworker and the caregiver (Fig. 1). While this toolallowed cleaning the variable and avoiding double counting a change in caregiver, it was also possible to graphically observethat the caseworker was often unaware of children’s moves, even if the case was open. Fig. 1 illustrates this concept usinga child that had four changes in households to other kin arrangements that were reported by the caregiver but not by theCWS.

Age at first change of caregiver. Given that newborn infants and those placed up to 3 months of age are likely to be adopted andreach a stable placement, we derived a variable based on the date of birth of the child and the dates provided for each changeas well as variable that represented the age of the infant at the first known change of caregiver/household. We categorizedage as 0–3 months or 4–12 months.

Substantiation. Substantiation is CPS’s statement about the legal status (proved findings) of a report of child maltreatment(Drake, 1996). We classified children on the basis of caseworkers’ responses as having a case substantiated or indicated(yes/no).

Levels of harm and risk. Caseworkers were queried about harm and risk regardless of the outcome of the investigation.Response categories were none, mild, moderate, and severe.

Maltreatment characteristics. NSCAW used the Limited Maltreatment Classification System (L-MCS) to capture informationabout the reported maltreatment from the caseworker. The most serious type of maltreatment was categorized into physicalabuse, neglect (failure to provide supervision or lack of supervision), and other.

Caseworker report on caregiver problems. NSCAW used the risk assessment questions from the risk assessment tools used inChild Protective Services in Michigan, New York, Washington, Illinois, and Colorado to collect information about the primarycaregiver. This tool includes 24 items; an example is “Did caregiver have any serious mental health or emotional problem?”

C. Casanueva et al. / Child Abuse & Neglect 38 (2014) 498–509 503

Table 2Caregiver instability during the first years of life.

Changes when child was 0 Changes 1 Change 2 or more changes

% SE % SE % SE

0–6 months of age 66.8 3.0 22.6 2.0 10.6 2.07–12 months of age 74.1 2.9 21.5 2.9 4.5 0.813–18 months of age 75.3 2.6 20.8 2.4 3.9 0.919–24 months of age 54.9 2.6 39.3 2.6 5.7 1.1

Total 0–24 months of age 14.4 2.0 34.6 3.3 51.0 3.325–30 months of age 78.2 3.0 19.6 3.1 27.0 0.731–36 months of age 94.1 1.1 5.5 1.0 0.4 0.237–42 months of age 92.4 2.8 6.9 2.8 0.7 0.443–48 months of age 84.4 1.7 15.2 1.7 0.4 0.249–54 months of age 74.4 2.0 25.2 2.0 0.4 0.255–60 months of age 79.2 3.2 19.4 2.9 1.3 0.6

N

Ab

Sfc

CChl

Bda

Ccp

Co

RmtmsospwaI(

0f

wdie

61–66 months of age 93.5 1.1 6.2 1.1 0.3 0.2Total study period 4.6 1.1 9.7 1.7 85.7 2.1

ote. All percentages are weighted, Ns are unweighted.

ny history of previous CPS involvement. Determination of previous CPS involvement was based on several questions ataseline to caseworkers; an example is, “Were there any prior reports of maltreatment to the agency?”

etting at baseline. Setting at baseline (in-home care, informal kin care, formal kin care, foster care, group care) was obtainedrom the CWS agency workers and completed with information from caregivers if there was not complete information fromaseworkers.

hild living skills. NSCAW used the Daily Living Skills items from the Vineland Adaptive Behavior Scale Screener (Sparrow,arter, et al., 1993). These items assess personal care skills, such as how the child eats, dresses, and performs personalygiene. Standardized scores were classified as Normal (scores ≥ 80), Moderate problems (scores = 71–79), or Severe prob-

ems (scores ≤ 70).

ayley infant neurodevelopmental screener (BINS). BINS is a screening tool to identify infants between 3 and 24 months old withevelopmental delays or neurological impairments for further diagnostic testing (Aylward, 1995). Children were classifieds high, moderate, or low risk for neurodevelopmental delay based on standardized scores.

hild chronic health condition. Caregivers were asked whether their child had any health problems that “last a long time orome back again and again.” The list of chronic conditions reported by caregivers included AIDS, brain tumor, and cerebralalsy.

hild physical disability. Caregivers were asked whether their child had any physical disability, including deafness, hearing,rthopedic, and visual impairment.

isk factor indices. To create the NSCAW risk index, two sources of information were used: the caseworkers’ report on theain caregivers’ risk factors and caregivers’ report. Most risk indicators concerned the primary caregiver at the time of

he index maltreatment, including victim of domestic violence, active abuse of substances like alcohol and/or illegal drugs,ental health problems, childhood history of abuse or neglect, poor parenting skills, arrest for any offense, incomplete high

chool education, and teen parent. A second set of risk factors included family instability and poverty indicators, such as fourr more children in the household, use of homeless shelter, low social support, receipt of child support payments or incomeupport by anyone in the household, difficulty paying for basic necessities, and high stress in the family. One risk indicatorertained to stressful situations for the child, such as hospitalized overnight for an injury or illness. For the caregiver, alongith interview questions, two instruments were used to determine risk factors. Maternal major depression, alcohol use,

nd drug use were assessed with the screening scales of the World Health Organization Composite International Diagnosticnterview Short-Form (CIDI-SF; Kessler, Andrews, et al., 1998). The physical violence subscale of the Conflict Tactics ScalesStraus, 1979) was used to assess caregivers’ experiences with physical domestic violence from an intimate partner.

The final list of risk factors is presented in Table 3. A total risk index score was generated by scoring each risk factor as(not present) or 1 (present) and summing them (range: 0–14). About a quarter of children (27.6%) had a risk index score

rom 0 to 3, 28.7% had a risk score of 4 or 5, 26.6% had a score of 6 or 7, and 17.2% had a score of 8 or more.The ACEs index includes psychological maltreatment, physical abuse, sexual abuse, and child neglect based on case-

orkers’ report. Indicators based on both caseworker and caregiver reports were caregiver alcohol or drug use, caregiverepression or mental health problems, caregiver treated violently, and criminal behavior in the household. A total ACEs

ndex score was generated by scoring each risk factor as 0 (not present) or 1 (present) and summing them (range: 0–8; Andat al., 2006).

504 C. Casanueva et al. / Child Abuse & Neglect 38 (2014) 498–509

Table 3NSCAW risk index indicators and caregiver instability.

Indicator at Baseline Total Incidence rate ratio

% SE IRR 95% CI p

Current victim of domestic violencea 43.6 3.2 1.13 1.03 1.25 0.01Use of alcohol and/or illegal drugsa 44.8 3.1 1.15 1.03 1.29 0.01Mental health problems/Depressiona 37.6 3.2 1.26 1.14 1.39 0.00Caregiver childhood history of abuse or neglectb 33.0 2.2 1.18 1.04 1.34 0.01Caregiver poor parenting skillsb 43.2 2.7 1.37 1.22 1.55 0.00Caregiver ever been arresteda 37.6 2.9 1.08 0.96 1.22 0.21Child hospitalized overnightc 20.0 2.3 1.14 1.00 1.30 0.06Teen parentc 26.9 3.3 1.22 1.07 1.39 0.00Incomplete high school educationc 30.5 2.3 1.05 0.93 1.17 0.44Four or more children in the housec 18.7 2.0 1.14 1.00 1.30 0.05Use of homeless shelterc 8.3 1.5 1.33 1.12 1.58 0.00Family has low social supporta 34.3 2.7 1.05 0.91 1.21 0.50Receipt of child or income support payments or difficulty paying for basic necessitiesa 86.1 1.8 1.14 0.90 1.45 0.27High stress in the familyb 60.8 2.7 1.16 1.00 1.35 0.05

Note. All percentages are weighted. p values represented as “0.00” were less than .001.

a Based on caregiver or caseworker report.b Based on caseworker report.c Based on caregiver report.

Analyses

We conducted the preliminary analyses with weighted data, using the SUDAAN statistical package version 9.0.1 (RTIInternational, 2007) to take into account NSCAW’s complex sampling design. All percentages are adjusted (weighted) foroversampling probabilities. Tables that include listed sample sizes have not been adjusted (i.e., are unweighted) in order toinform the reader of the subsample number of subjects; however, those values cannot be used directly to obtain percentages.Analyses examined predictors of children’s caregiver instability. Incidence rate ratios are used to report on the bivariateassociation between risk index indicators and caregiver instability. Negative binomial regression analyses modeled caregiverinstability as a function of child and caregiver characteristics, and risk indicators at baseline were fit using Mplus 7 (Muthén& Muthén, 2012). Adjusted incidence rate ratios (IRRs) with 95% confidence intervals in the negative binomial regressionmodels and p values based on Wald tests are reported. Prior to fitting the negative binomial regression models, 60 imputeddata sets were generated to address problems of missing data, and estimates are aggregated over results from fitting themodels (Rubin, 1976). Multiple imputation has been shown to reduce bias in parameter estimates and increase statisticalpower relative to complete-case analyses (Enders, 2010, 2012). Given that some of the covariates used in the models overlapwith some of the NSCAW Index and the ACEs index, the overlapping covariates were excluded from those models.

Results

Experience of changing caregivers among maltreated infants

Over the course of the study, 37.8% of the children were placed in out-of-home care as reported by CWS caseworkers.Nevertheless, reported out-of-home placements accounted for only a portion of caregiver changes when information fromboth caregivers and caseworker was integrated across time. A change in caregiver was very common in the first 2 years oflife for infants reported to the CWS for maltreatment. During the first 6 months of life, 33.2% of children experienced at leastone change, whereas during the second 6 months of life, 25.9% experienced at least one change. From 13 to 18 months old,24.7% of children experienced at least one change, and almost half of the children (45.1%) experienced at least one changefrom 19 to 24 months. Overall, 85.6% of children experienced one or more changes during the first 2 years of life. More thanhalf (51.0%) experienced two or more changes (Table 2). From infancy to 5–7 years old, 95.4% had at least one change; withinthis group, 9.7% had one change, 47.1% had two or three changes, 25.5% had four or five changes, and 13.2% had six or morechanges. The mean number of changes was 3.5 (SD = 2.1, range: 0–19). In bivariate and multiple predictor models, analysis ofchanges by children’s gender, race/ethnicity, placement in foster care, and substantiation, showed no significant differencesamong subgroups of children.

Association between risk factors and caregiving instability

The mean number of risk factors in the NSCAW risk index at the time of the index report of maltreatment (baseline)was 5.1 (range: 0–12). Children with higher NSCAW risk indices were more likely to have multiple caregiver changes thanwere children with lower NSCAW risk indices. For children with three or fewer risk factors, 62.7% experienced two or threechanges. Almost half (43.1%) of the children with six or seven risk factors had four or more changes, and 71.5% of children

C. Casanueva et al. / Child Abuse & Neglect 38 (2014) 498–509 505

Table 4Results of negative binomial regression models predicting caregiver instability.

Child characteristic Model 1 Model 2 Model 3

IRR 95%CI p IRR 95%CI p IRR 95%CI p

Gender (ref. male)Female 0.93 (0.84, 1.03) 0.183 0.95 (0.86, 1.05) 0.321 0.95 (0.86, 1.05) 0.303

Race/ethnicity (ref. white)Black 0.90 (0.8, 0.99) 0.052 0.92 (0.83, 1.04) 0.177 0.93 (0.83, 1.04) 0.222Hispanic 0.99 (0.87, 1.13) 0.902 1.00 (0.87, 1.15) 0.972 1.00 (0.88, 1.14) 0.985Other 1.10 (0.93, 1.30) 0.25 1.12 (0.95, 1.33) 0.18 1.14 (0.99, 1.31) 0.077

Setting at baseline (ref. in-home care)Informal kin care 1.11 (0.88, 1.40) 0.361 1.13 (0.90, 1.42) 0.285 1.22 (1.02, 1.46) 0.028Formal kin care 1.12 (0.84, 1.48) 0.442 1.12 (0.84, 1.50) 0.434 1.23 (0.94, 1.62) 0.13Foster care 1.08 (0.96, 1.23) 0.215 1.06 (0.93, 1.20) 0.384 1.09 (0.97, 1.24) 0.154Group home care 1.23 (0.94, 1.61) 0.124 1.19 (0.90, 1.57) 0.212 1.22 (0.93, 1.59) 0.144

Daily living skills (ref. moderate to severe problems)Normal 0.83 (0.58, 1.20) 0.331 0.84 (0.57, 1.24) 0.384 0.88 (0.60, 1.29) 0.5

Risk of neurodevelopmental delay (ref. high risk)Low or no risk 1.04 (0.88, 1.23) 0.658 1.01 (0.86, 1.19) 0.901 0.99 (0.83, 1.18) 0.934Medium risk 0.95 (0.86, 1.04) 0.239 0.93 (0.85, 1.02) 0.141 0.92 (0.84, 1.02) 0.097

Chronic health condition (ref. No)Yes 1.13 (1.01, 1.26) 0.038 1.12 (0.99, 1.26) 0.062 1.10 (0.98, 1.25) 0.113

Physical disability (ref. no)Yes 1.03 (0.85, 1.25) 0.743 1.03 (0.84, 1.26) 0.774 1.06 (0.87, 1.30) 0.551

Age at first change of caregiver (ref. 4 months of age or more)0–3 months of age 0.75 (0.67, 0.84) 0.000 0.76 (0.69, 0.85) 0.000 0.77 (0.69, 0.86) 0.000

Substantiation (ref. no)Substantiated/indicated 1.09 (0.97, 1.22) 0.15 1.08 (0.96, 1.21) 0.203 1.04 (0.94, 1.15) 0.417

Any history of previous CPS involvement (ref. no)Yes 1.08 (0.99, 1.18) 0.093 1.03 (0.93, 1.13) 0.581 1.03 (0.93, 1.13) 0.624

Main type of maltreatment (ref. other)Physical 0.77 (0.65, 0.90) 0.002 0.77 (0.65, 0.92) 0.003Neglect 0.99 (0.84, 1.16) 0.895 1.00 (0.85, 1.17) 0.968

Caregiver characteristicsAge (ref. age 20–39)Age 19 or younger 1.13 (0.98, 1.31) 0.105 1.12 (0.97, 1.3) 0.128Age 40 or older 1.3 (1.12, 1.52) 0.001 1.27 (1.10, 1.47) 0.001

Education (ref. less than high school)High school 0.9 (0.81, 0.99) 0.048 0.92 (0.83, 1.03) 0.142More than high school 0.94 (0.81, 1.09) 0.413 0.94 (0.81, 1.09) 0.421

Marital status (ref. not married)Married 0.93 (0.82, 1.04) 0.199 1.00 (0.89, 1.11) 0.959

Risk index 1.06 (1.04, 1.09) 0.000ACEs 1.06 (1.03, 1.1) 0.000

NrB

ww

rtw1n

P

ipac4oc

ote. IRR = incidence rate ratio. CI = limits of the 95% confidence interval. IRRs are obtained using negative binomial regression, a generalization of Poissonegression used when the variance is larger than the mean of a count variable, a condition known as overdispersion.old indicates statistically significant.

ith eight or more risk factors had four or more changes. All comparisons between fewer risk factors and higher risk factorsere statistically significant, showing that more risk factors were associated with higher caregiver instability.

The percentage of children exposed to each risk factor included in the NSCAW risk index and the association between theisk factor with caregiver instability based on bivariate analysis are presented in Table 3. IRRs presented in Table 3 representhe percent increase in expected count of changes during the study period. Although most of the NSCAW risk index variablesere significantly associated with increases in caregiver instability, the expected count of change increase was only between

3% and 37% for each individual variable. For example, the effect of having a caregiver with a childhood history of abuse oreglect increased the expected count of changes by 18% (IRR: 1.18, p = .01).

redictors of caregiver instability among infants reported to the CWS

We used several models to predict the count of caregiver changes. In the first model, predictors of changes in caregiversdentified in previous studies were analyzed. Five predictors were associated with changes in caregivers; two of theseredictors increased the rate of change, whereas three predictors decreased the rate of change. IRRs are presented in Table 4nd the values represent the effect of each variable holding all other variables constant at their mean. Children having a

hronic health condition had an increase in the expected count of changes of 13%; children with a caregiver older than0 years of age at baseline increased the expected count of changes by 30%. In contrast, children who had their first changef caregiver between 0 and 3 months compared with 4 months or older decreased the expected count of changes by 25%;hildren who were reported for physical abuse compared with “other” types of abuse decreased the expected count of

506 C. Casanueva et al. / Child Abuse & Neglect 38 (2014) 498–509

changes by 23%; and children of caregivers who had a high school education compared with those with less than a highschool education decreased the expected count of changes by 10%.

The second model includes predictors identified in previous studies, and the ACEs index. For each unit of increase of theACEs index, the increase in changes was 6%. A model using only the ACEs index as predictor (no covariates) showed that foreach unit of increase of the ACEs index, the increase in the expected count of change was 9%.

We used a similar strategy for the third model but using the NSCAW risk index. For each unit of increase of the NSCAWrisk index, the increase in the expected count of changes was 6%. A model using only the NSCAW risk index (no covariates)as predictor showed that for each unit of increase of the NSCAW risk index the increase in changes was 8%.

Note that the IRR for continuous predictors is interpreted differently than for binary predictors. For each additional unit inthe risk indices, there is an increase of 6–9% (depending on the model and index). For example, increasing a child’s ACEs 311count by 2 would increase the expected counts of changes to 1.092 = 1.19 or an 18% increase. An increase to the maximumof the ACEs (8 points) increases the expected counts of changes in caregivers to 1.99 or a 99% increase. As the indices arethemselves count variables, having the highest risk index dramatically moves the expected count of changes in caregivers.

Discussion

Overall, 85.6% of children who were infants at the time of the index maltreatment experienced at least one changeof caregiver and/or household during their first 2 years of life. Almost 40% of children experienced four or more changesbetween infancy and entering school. Changes of caregiver and/or household so early in life can have profound implicationsfor attachment and child development. Importantly, all infants who were investigated for a report of maltreatment wereat high risk for instability, regardless of the substantiation status of the maltreatment report, whether the child remainedin-home or was placed out of home, or the child’s race/ethnicity or gender.

These findings challenge the common belief that the CWS is the source of all patterns of caregiver instability observed inmaltreated infants. Instability may possibly be one aspect of a high-risk environment that ultimately draws CWS attention.This study primarily identifies a high risk for instability of care among all infants investigated, both those placed in out-of-home care and those who were never placed. Regardless of whether CWS involvement contributes to instability, contactwith CWS provides an important source for identifying infants at risk for instability of care.

This study shows that instability of caregiver is a more general risk for children who come into contact with CWS,independent of substantiation and out-of-home placement. This risk raises the questions: Could the CWS play a moreprominent preventive role? How should the CSW be involved in young children’s lives to promote stability for those left athome after a maltreatment investigation, given that they may no longer be formally involved in the CWS? In light of theextremely high levels of caregiver instability reported in this study among all infants, these questions are paramount. As apoint of comparison, in the Early Head Start Research and Evaluation Project, the vast majority (89%) of families were livingbelow the poverty line, but only 16% of children had a change of caregiver of a week or longer during the first 2 years of life,and separation was rarely caused by the child being removed from the home by the CWS. Early mother–child separation ofa week or longer within the first 2 years of Early Head Start children’s lives was related to higher levels of child negativitytoward the mother (e.g., anger, hostility, or dislike toward the mother) at age 3 and aggression at 3 and 5 years old (Howardet al., 2011).

It is noteworthy that, even among children who had an open case with the CWS across time, caseworkers were not able toreport all the changes in caregiver/household described by caregivers. This explains the high estimate of caregiver instabilityin this study compared with previous studies of placement instability based on CWS information (Horwitz et al., 2011;Wulczyn et al., 2003). Children with reports of maltreatment that were not substantiated had the same level of caregiverinstability as children with substantiated reports, typically moving from one relative’s household to the next several timesduring early childhood. The mean number of changes among this vulnerable group of infants was 3.5, with some reaching19 changes by the time of school entry. The operationalization of caregiver instability as an integration of caseworker andcaregiver data is relevant for future studies. It is possible that previous studies have been constrained by their use of onlycaseworker data or administrative data in assessing the impact on children (Harden & Whittaker, 2011).

Some predictors of placement reported in previous studies were significantly associated with caregiver instability inthis study. Higher caregiver instability was associated with children having a chronic health condition, a risk factor previ-ously reported as being associated with out-of-home placement in general and with multiple placements (Bhatti-Sinclair &Sutcliffe, 2012; Eggertsen, 2008). This finding highlights the many challenges that caregivers face when taking care of youngchildren with chronic health problems, with those problems potentially exhausting the care capacity of the family or care-giver. In those cases, additional support and services, including access to Medicaid, may help prevent caregiver instability.Having a caregiver older than age 40 at baseline also increased the expected count of changes, as described in a previousstudy (Berger et al., 2009). Older caregivers, both kin and biological mothers, may have less energy and more difficulty takingcare of very young children. As with children with health care problems, a very active infant or toddler can also exhaust thecare capacity of the caregiver, and indicate the need for support services when the care of young children is left to older

caregivers.

Covariates associated with greater stability include having the first change of caregiver between 0 and 3 months of age(relative to experiencing the change at more than 3 months of age), a finding described in previous studies in association withvery young infants being rapidly adopted (Wulczyn et al., 2002). Foster parents report feeling more committed to infants

paatgfiral

at(r

phwmwoit

L

icatdebtd

R

Fntie

radftmTtf2ps(o

C. Casanueva et al. / Child Abuse & Neglect 38 (2014) 498–509 507

laced at younger ages, which could contribute to this greater caregiver stability (Dozier & Lindhiem, 2006). The finding thatreport of physical abuse was associated with less caregiver instability may also indicate that, for these youngest infants,swift and definitive resolution tends to be taken by the CWS. Infants are extremely fragile and physical abuse can put

heir lives at risk. If the court decision is to terminate parental rights and the infant can be adopted or a relative can haveuardianship, those placements could be permanent, explaining the greater stability associated with physical abuse. Thending that infants with a caregiver that has at least a high school education at baseline experience less instability has beeneported previously (Berger et al., 2009). This finding may result from mothers with higher educational levels having greaterccess to resources that can facilitate the daily task of taking care of an infant or young child as compared with mothers withower educational levels.

Of the two risk indices used in this study, both have a similar predictive capacity and demonstrate that counts of risksre more predictive of instability than are individual risk factors, either if the index focuses mostly in the experience of mal-reatment (the case of the ACEs index) or if the index focuses on the risk factors surrounding the experience of maltreatmentthe case of the NSCAW risk index). These indices demonstrate that beyond the impact of any given risk factor, cumulativeisk indices predict caregiver instability in powerful ways for young children.

Most infants reported to the CWS for maltreatment described in this study were exposed to multiple risk factors, such ashysical or emotional abuse, neglect, caregiver substance abuse or mental illness, exposure to violence, and family economicardship. Children with a higher number of risk factors were significantly more likely to have caregiver instability at an agehen having a stable caregiver is critical for the child’s well-being and development. The repeated loss of a young child’sain caregiver or the experience of caregivers who are unavailable or neglectful are likely experienced as traumatic events,hich places children at risk for numerous adverse psychological sequelae later in development (National Scientific Council

n the Developing Child, 2005). Research has also suggested that adults with a history of recurring childhood trauma are atncreased risk for emotion regulation difficulties and vulnerability to develop traumatic symptoms when exposed to newraumatic events (Ehring & Quack, 2010).

imitations

Several limitations of this study should be noted. First, the data were derived from caregivers’ self-reports, caseworkernterviews, and case records. Separate measures to triangulate on constructs were not available. Administrative files andaseworkers’ knowledge of a family may be incomplete, leading to significant underreporting of important variables, suchs family risk factors. Similarly, social response bias could affect reports from biological parents. Second, we did not examinehe differential impact of a type of caregiver/household on the next placement move, and there may be differential effectsepending on the order of types of care. Third, the variables used in the present study did not fully mirror other studiesxamining similar issues that exclusively use official reports of changes in placement by the CWS. This limits the compara-ility with previous studies. However, the present study made use of all available waves of infants’ data, and we includedhe description of two critical reporters. The integration of this information provides a much more detailed and worrisomeescription of caregiver instability among infants than any previous study.

esearch and policy implications

The impact of maltreatment and caregiver instability has been recognized in federal legislation. The 2011 Child andamily Services Improvement and Innovation Act (P.L. 112-34, reauthorizing Title IV-B of the Social Security Act) includesew language that requires states to identify and develop mental health oversight plans to “monitor and treat emotionalrauma associated with a child’s maltreatment and removal” (p. 2; DHHS, 2011). The new legislation supports the CWS toncorporate effective interventions for maltreated children experiencing toxic stress and trauma (Samuels, 2011) and makefforts for the first placement of infants to be the last placement (Zero To Three, 2009).

Some evidence-based programs are designed to work with children experiencing adversity related to maltreatment,emoval, and abandonment. These interventions include Attachment and Biobehavioral Catch-up (ABC), a home-basedpproach that targets foster and high-risk birth children’s dysregulation and challenging behaviors, helping parents of chil-ren aged birth to 3 years old provide nurturing, sensitive care that promotes child regulatory capabilities and attachmentormation (Dozier, Peloso, et al., 2006); the Bucharest Early Intervention Project, developed for institutionalised infants andoddlers in Bucharest, Romania, that includes specialized support for foster parents concerning infant mental health, attach-

ent development, and management of behavioral and emotional problems (Zeanah, Nelson, et al., 2003); Multidimensionalreatment Foster Care for Preschoolers (MTFC) for children 3–6 years old, a family-based intervention directed at children, fos-er care providers, and permanent caregivers that includes intensive foster parent training and daily support, child servicesrom a behavioral specialist, family therapy, and medication management if necessary (Fisher, Gunnar, Dozier, Bruce, & Pears,006); and Child Parent Psychotherapy, a high-intensity (50 or more 1-h weekly sessions) attachment- and trauma-focused

sychotherapeutic intervention directed at young children and parents together and focusing on child–parent relation-hip (Lieberman, Ippen, & Van Horn, 2006). These evidence-based approaches have been shown to improve attachmentBernard, Dozier, et al., 2012), child well-being outcomes (Fisher et al., 2006; Lieberman et al., 2006), and child welfareutcomes (Fisher, Kim et al., 2009) for children involved with the CWS.

508 C. Casanueva et al. / Child Abuse & Neglect 38 (2014) 498–509

The data presented in the present study describe the vulnerabilities among infants reported for maltreatment. The extentof the early experience of caregiver instability is several folds greater than that of the population of children of similar lowsocioeconomic level. The number of caregiver/household changes that these young children in the CWS experience is veryhigh, with more than half of children experiencing two or more changes. The profile of family characteristics provided inthe study provides a portrait of disadvantages that leaves children at high risk for negative developmental outcomes andalso at higher risk of experiencing further caregiver instability. These data underscore the importance of national efforts toimprove social-emotional outcomes for children involved in the CWS. Child welfare agencies are encouraged to recognizethe risk that caregiving instability presents for developing children, to promote case planning that recognizes infants’ andyoung children’s need for consistent and sensitive caregiving, and to incorporate and implement programs and services thatare both stress/trauma-informed as well as evidence-based (DHHS, 2012a).

Acknowledgements

We would like to express thanks for the expert editorial assistance provided by Jeff Novey and Laura Small.

References

Ainsworth, M., Blehar, M., Waters, E., & Wall, S. (1978). Patterns of attachment: A psychological study of the strange situation. Hillsdale, NJ: Erlbaum.Anda, R., Felitti, V., Bremner, J., Walker, J., Whitfield, C., Perry, B., et al. (2006). The enduring effects of abuse and related adverse experiences in childhood:

A convergence of evidence from neurobiology and epidemiology. European Archives of Psychiatry and Clinical Neuroscience, 256(3), 174–186.Aylward, G. P. (1995). Bayley infant neurodevelopmental screener. San Antonio, TX: Psychological Corporation.Berger, L., Bruch, S., Johnson, James, & Rubin. (2009). “Estimating the impact” of out-of-home placement on child well-being: Approaching the problem of

selection bias. Child Development, 80(6), 1856–1876.Bernard, K., Dozier, M., Bick, J., Lewis-Morrarty, E., Lindhiem, O., & Carlson, E. (2012). Enhancing attachment organization among maltreated children:

Results of a randomized clinical trial. Child Development, 83(2), 623–636.Bhatti-Sinclair, K., & Sutcliffe, C. (2012). What determines the out-of-home placement of children in the USA? Children and Youth Services Review, 34(9),

1749–1755.Biemer, P., Dowd, K., & Webb. (2010). Study design and methods. In M. Webb, K. Dowd, B. Harden, M. Landsverk, & Testa (Eds.), Child welfare and child

well-being: New perspectives from the National Survey of Child and Adolescent Well-Being. New York: Oxford University Press.Dalaker, J. (2001). Poverty in the United States: 2000. Retrieved from http://ict.cas.psu.edu/resources/Census/PDF/C2K Poverty in USA.pdfDozier, M., & Lindhiem, O. (2006). This is my child: Differences among foster parents in commitment to their young children. Child Maltreatment, 11(4),

338–345.Dozier, M., Peloso, E., Lindhiem, O., Gordon, M. K., Manni, M., Sepulveda, S., et al. (2006). Developing evidence-based interventions for foster children: An

example of a randomized clinical trial with infants and toddlers. Journal of Social Issues, 62(4), 767–785.Drake, B. (1996). Unraveling unsubstantiated. Child Maltreatment, 1(3), 261–271.Eggertsen, L. (2008). Primary factors related to multiple placements for children in out-of-home care. Child Welfare, 87(6), 71–90.Ehring, T., & Quack, D. (2010). Emotion regulation difficulties in trauma survivors: The role of trauma type and PTSD symptom severity. Behavior Therapy,

41(4), 587–598.Enders, C. K. (2010). Applied missing data analysis. New York, NY: The Guilford Press.Enders, C. K. (2012). Dealing with missing data in developmental research. Child Development Perspectives, 7(1), 27–31.Fearon, R. P., Bakermans-Kranenburg, M. J., van Ijzendoorn, M. H., Lapsley, A. M., & Roisman, G. I. (2010). The significance of insecure attachment and

disorganization in the development of children’s externalizing behavior: A meta-analytic study. Child Development, 81(2), 435–456.Fisher, P., Gunnar, M., Dozier, Bruce, & Pears. (2006). Effects of therapeutic interventions for foster children on behavioral problems, caregiver attachment,

and stress regulatory neural systems. Resilience in Children, 1094, 215–225 (B. Lester, A. Masten and B. McEwen)Fisher, P., Kim, H., & Pears, K. (2009). Effects of Multidimensional Treatment Foster Care for Preschoolers (MTFC-P) on reducing permanent placement

failures among children with placement instability. Children and Youth Services Review, 31(5), 541–546.Foster, E., Hillemeier, M., & Bai. (2011). Explaining the disparity in placement instability among African-American and white children in child welfare: A

Blinder-Oaxaca decomposition. Children and Youth Services Review, 33(1), 118–125.Goerge, R., & Harden, A. (1993). The impact of substance-exposed infants on child protective services and substitute care caseloads: 1985–1992. In A report

to the Illinois Department of Children and Family Services. Chicago: Chapin Hall Center for Children.Harden, B., & Whittaker, J. (2011). The early home environment and developmental outcomes for young children in the child welfare system. Children and

Youth Services Review, 33(8), 1392–1403.Horwitz, S., Hurlburt, M., Cohen, Zhang, & Landsverk. (2011). Predictors of placement for children who initially remained in their homes after an investigation

for abuse or neglect. Child Abuse & Neglect, 35(3), 188–198.Howard, K., Martin, A., Berlin, & Brooks-Gunn. (2011). Early mother–child separation, parenting, and child well-being in Early Head Start families. Attachment

& Human Development, 13(1), 5–26.Kessler, R. C., Andrews, G., Mroczek, D., Ustun, T. B., & Wittchen, H.-U. (1998). The World Health Organization Composite International Diagnostic Interview

Short Form (CIDI-SF). International Journal of Methods in Psychiatric Research, 7, 171–185.Lewis, E., Dozier, M., Ackerman, J., & Sepulveda-Kozakowski, S. (2007). The effect of placement instability on adopted children’s inhibitory control abilities

and oppositional behavior. Developmental Psychology, 43(6), 1415–1427.Lieberman, A., Chu, A., Van Horn, P., & Harris, W. (2011). Trauma in early childhood: Empirical evidence and clinical implications. Development and

Psychopathology, 23, 397–410.Lieberman, A., Ippen, C., & Van Horn. (2006). Child–parent psychotherapy: 6-Month follow-up of a randomized control trial. Journal of the American Academy

of Child and Adolescent Psychiatry, 45(8), 913–918.Lightfoot, E., Hill, K., & LaLiberte, T. (2011). Prevalence of children with disabilities in the child welfare system and out of home placement: An examination

of administrative records. Children and Youth Services Review, 33(11), 2069–2075.Muthén, L., & Muthén, B. (2012). Mplus User’s Guide. 7th ed. Los Angeles, CA.National Scientific Council on the Developing Child. (2005). Excessive Stress Disrupts the Architecture of the Developing Brain. From

www.developingchild.harvard.eduRTI International. (2007). SUDAAN user’s manual, release 9.0.1. NC: Research Triangle Park.Rubin, D. (1976). Inference and missing data. Biometrika, 63(3), 581–592.Rubin, D., O‘Reilly, A., Luan, & Localio. (2007). The impact of placement stability on behavioral well-being for children in foster care. Pediatrics, 119(2),

336–344.

S

S

S

SU

U

U

UWWWZ

Z

Z

C. Casanueva et al. / Child Abuse & Neglect 38 (2014) 498–509 509

amuels, B. H. (2011). Addressing trauma to promote social and emotional well-being: A child welfare imperative. In J. Collins, K. Decker, & S. HKo (Eds.),Effectively addressing the impact of child traumatic stress in child welfare. Arlington, VA: CWLA.

immel, C., Morton, C., & Cucinotta. (2012). Understanding extended involvement with the child welfare system. Children and Youth Services Review, 34(9),1974–1981.

parrow, S. S., Carter, A. S., & Cicchetti, D. V. (1993). Vineland screener: Overview, reliability, validity, administration, and scoring. New Haven, CT: Yale UniversityChild Study Center.

traus, M. A. (1979). Measuring intrafamily conflict and violence: The Conflict Tactics (CT) Scale. Journal of Marriage and the Family, 41, 75–88..S. Department of Health and Human Services, Administration for Children and Families. (2011). Information memorandum ACYF-CB-IM-11-06. Washington,

DC: Children’s Bureau..S. Department of Health and Human Services, Administration for Children and Families. (2012). Promoting Social and Emotional Well-Being for Children

and Youth Receiving Child Welfare Services. Retrieved from http://www.acf.hhs.gov/programs/cb/laws policies/policy/im/2012/im1204.pdf.S. Department of Health and Human Services. (2012). The AFCARS report # 19. Retrieved from http://www.acf.hhs.gov/sites/default/files/main/

afcarsreport19.pdf.S. Department of Health and Human Services. (2012). Child maltreatment 2011. Retrieved from http://www.acf.hhs.gov/sites/default/files/cb/cm11.pdfebster, D., Barth, R. P., & Needell. (2000). Placement stability for children in out-of-home care: A longitudinal analysis. Child Welfare, 79(5), 614–632.ulczyn, F., Hislop, K. B., & Harden. (2002). The placement of infants in foster care. Infant Mental Health Journal, 23(5), 454–475.ulczyn, F., Kogan, J., & Harden, B. J. (2003). Placement stability and movement trajectories. Social Service Review, 77(2), 212–236.

eanah, C., Berlin, L., & Boris, N. (2011). Practitioner review: Clinical applications of attachment theory and research for infants and young children. Journalof Child Psychology and Psychiatry, 52(8), 819–833.

eanah, C., Nelson, C., Fox, N., Smyke, A., Marshall, P., Parker, S., et al. (2003). Designing research to study the effects of institutionalization on brain andbehavioral development. The Bucharest Early Intervention Project Development and Psychopathology, 15(4), 885–907.

ERO TO THREE. (2009). Infants and toddlers in foster care. From http://zerotothree.org/