Development and Psychopathology 18 ~2006!, 623–649 Copyright © 2006 Cambridge University Press Printed in the United States of America DOI: 10.10170S0954579406060329

Fostering secure attachment in infants in maltreating families through preventive interventions

DANTE CICCHETTI,a FRED A. ROGOSCH,b and SHEREE L. TOTHb aInstitute of Child Development, University of Minnesota; and bMt. Hope Family Center, University of Rochester

Abstract The malleability of insecure and disorganized attachment among infants from maltreating families was investigated through a randomized preventive intervention trial. Findings from research on the effects of maltreatment on infant attachment were incorporated into the design and evaluation of the intervention. One-year-old infants from maltreating families ~N ϭ 137! and their mothers were randomly assigned to one of three intervention conditions: ~a! infant–parent psychotherapy ~IPP!, ~b! psychoeducational parenting intervention ~PPI!, and ~c! community standard ~CS! controls. A fourth group of infants from nonmaltreating families ~N ϭ 52! and their mothers served as an additional low-income normative comparison ~NC! group. At baseline, mothers in the maltreatment group, relative to the nonmaltreatment group mothers, reported greater abuse and neglect in their own childhoods, more insecure relationships with their own mothers, more maladaptive parenting attitudes, more parenting stress, and lower family support, and they were observed to evince lower maternal sensitivity. Infants in the maltreatment groups had significantly higher rates of disorganized attachment than infants in the NC group. At postintervention follow-up at age 26 months, children in the IPP and PPI groups demonstrated substantial increases in secure attachment, whereas increases in secure attachment were not found for the CS and NC groups. Moreover, disorganized attachment continued to predominate in the CS group. These results were maintained when intent to treat analyses were conducted. The findings are discussed in terms of the utility of translating basic research into the design and evaluation of clinical trials, as well as the importance of preventive interventions for altering attachment organization and promoting an adaptive developmental course for infants in maltreating families.

An overarching goal of the field of prevention dation for prevention science ~Cicchetti & science is to intervene in the course of devel- Hinshaw, 2002; Cicchetti & Toth, 1992; Ia- opment to ameliorate or eliminate the emer- longo et al., 2006; Institute of Medicine, 1994!. gence of maladaptation and psychopathology An organizational perspective on devel- ~Ialongo et al., 2006!. The discipline of devel- opment has been an influential theoretical opmental psychopathology, with its major fo- approach in the field of developmental psy- cus on the dialectic between the study of chopathology ~Cicchetti, 1993; Cicchetti & normality and pathology, is in a unique posi- Schneider-Rosen, 1986; Cicchetti & Sroufe, tion to provide an important theoretical foun- 1978; Sroufe & Rutter, 1984!. Theorists who adhere to an organizational perspective direct prevention science to focus on the progressive This research was supported by grants from the Admin- organization of developmental competencies istration of Children, Youth, and Families, the National and incompetencies in the course of epigene- Institute of Mental Health ~MH54643!, and the Spunk sis with the goal of structuring preventive ef- Fund, Inc. Address correspondence and reprint requests to: Dante forts ~Cicchetti & Toth, 1992; Ialongo et al., Cicchetti, Institute of Child Development, University of 2006; Toth & Cicchetti, 1999!. To effect change Minnesota, 51 East River Road, Minneapolis, MN 55455. in the course of development and avert psy-

623 624 D. Cicchetti, F. A. Rogosch, and S. L. Toth chopathological outcomes, preventive inter- Child maltreatment poses severe risks for ventions informed by an organizational long-term maladjustment and the develop- perspective should focus on promoting com- ment of psychopathology ~Cicchetti & Toth, petence and reducing ineffective resolution of in press; Cicchetti & Valentino, 2006!. Child the stage-salient developmental issues that maltreatment exemplifies a pathogenic rela- emerge at different periods in ontogenesis. By tional environment that is far beyond the range adopting the approach emphasized by organi- of what is normatively encountered and en- zational theorists, the deflection of maladap- genders substantial risk for maladaptation tation onto more adaptive developmental across diverse domains of biological and psy- pathways may be achieved, thereby enhanc- chological development ~Cicchetti & Manly, ing the individual’s capacity for a greater like- 2001; Cicchetti & Valentino, 2006; DeBellis, lihood of subsequent successful adaptation 2001!. Both the proximal environment of the ~Cicchetti, 1993; Cicchetti & Tucker, 1994; immediate family and the more distal fac- Sroufe, 1989!. Inherent in the organizational tors associated with the culture and the com- perspective is the importance of early inter- munity, as well as the transactions that occur vention, before developmental liabilities may among these ecological contexts, conspire to become consolidated ~Cicchetti, Toth, & Bush, undermine normal biological and psychologi- 1988; Toth & Cicchetti, 1999!. cal developmental processes in maltreated chil- Although the quintessential goal of preven- dren ~Cicchetti, 2002; Cicchetti & Lynch, 1993; tion science is to right the developmental DeBellis, 2001; Pollak, Cicchetti, Klorman, course and prevent the emergence of psycho- & Brumaghim, 1997!. Efforts to prevent the pathology, the results of prevention trials also deleterious consequences of maltreatment are possess important implications for develop- thus of critical importance. mental theory. As developmental experiments, In this paper, we report on the evaluation prevention trials provide a wealth of informa- of the efficacy of two theoretically informed, tion about the processes of typical and atypi- randomized preventive interventions for mal- cal development ~Cicchetti & Hinshaw, 2002; treating mothers and their 1-year-old infants. Cicchetti & Toth, in press; Hinshaw, 2002; Guided by an organizational perspective on Howe, Reiss, & Yuh, 2002; Ialongo et al., development, and by the translation of empir- 2006; Kellam & Rebok, 1992!. The transla- ical research that documents that maltreated tion of developmental theory into the design children manifest impairments in attachment and implementation of preventive interven- organization, in this investigation we inter- tions ~Cicchetti & Toth, 1992; Noam, 1992; vened early in the life course of maltreated in- Shirk, 1999; Toth & Cicchetti, 1999! and the fants to prevent the compromised developmental results of preventive interventions must form attainments that accompany maltreatment and a circular link back to the conceptual frame- that are precursors to later maladaptation and work to advance both the theory and future psychopathology. randomized intervention trials ~Cicchetti & The capacity for preferential attachment Hinshaw, 2002; Ialongo et al., 2006!. originates during early experiences and inter- If the developmental course is altered due to actions with the caregiver ~Gunnar & Vazquez, a preventive intervention, and the risk of the 2006; Schore, 2003; Sroufe, 1979, 1996!. The disorder or negative outcome is reduced, then preattachment parent–infant environment helps the research results will contribute to our under- to shape children’s physiological regulation standing of developmental processes ~Cic- and biobehavioral patterns of response ~Gun- chetti & Rogosch, 1999; Coie et al., 1993!. nar & Vazquez, 2006; Hofer, 1987; Pipp & Conversely, if the reduction of a targeted risk Harmon, 1987!. Near the end of the first year factor does not appear to have altered the patho- of life, normally developing infants derive feel- genic process, then that risk factor would not ings of security from their caregivers and use be viewed as a causal agent, but may be a marker them as a base from which to explore the of atypical development ~Cicchetti & Hinshaw, environment ~Cummings & Davies, 1996; 2002; Hinshaw, 2002; Kraemer et al., 1997!. Sroufe, 1979, 1996!. Dyadic interactions, Fostering secure attachment in infants 625 marked by relatedness and synchrony, resil- issue primes the way subsequent developmen- iency to stress, and appropriate affective inter- tal tasks are likely to be negotiated ~Cicchetti change, are associated with the development & Schneider-Rosen, 1986; Sroufe & Rutter, of a secure attachment relationship ~Sroufe, 1984!. Because maltreated infants predomi- 1979, 1996!. In the absence of regular contin- nantly develop insecure attachments, there is a gent responsivity, neither infant nor caregiver high probability that these babies will unsuc- derives feelings of efficacy in the relationship cessfully resolve subsequent developmental is- and the development of a secure attachment sues, paving the way for these infants to proceed is likely to be impeded ~Ainsworth, Blehar, along maladaptive developmental pathways. Waters, & Wall, 1978; Sroufe & Waters, The likelihood that maladaptation and psycho- 1977!. pathology will occur is exacerbated by the Because there are many enduring problem- preponderance of Type D attachments in the atic characteristics of the home environments maltreated babies. The developmental sequelae where maltreatment has occurred, such as emo- of disorganized–disoriented attachments are of- tional or physical rejection, aggression or hos- ten characterized by more maladaptive and tile management, threatening affective or verbal pathological outcomes than is the case for other assaults, or lack of appropriate responsivity types of attachment insecurity ~Carlson, 1998; ~Azar, 2002; Cerezo, 1997; Crittenden & Lyons-Ruth, Easterbrooks, & Cibelli, 1997; Ainsworth, 1989; Howes & Cicchetti, 1993; van IJzendoorn, Schuengel, & Bakermans- Rogosch, Cicchetti, Shields, & Toth, 1995!,as Kranenburg, 1999!. Thus, the preventive in- expected, research investigations have discov- terventions to promote the attainment or ered that the vast majority of the attachment maintenance of secure attachment organi- relationships between maltreated youngsters zation in maltreated infants are crucial for and their caregivers are pathological in na- redirecting the course of development onto adap- ture. The attachments of maltreated infants tive pathways. consistently have been shown to be insecure Sharing the common goal of preventing sub- ~Crittenden, 1988; Egeland & Sroufe, 1981; sequent inadequate parenting to avert mal- Schneider-Rosen, Braunwald, Carlson, & Cic- adjustment in maltreated infants, the two chetti, 1985!. Moreover, given the unpredict- interventions differed in their strategies for able, chaotic relationships that characterize the attaining these goals. The first model, psycho- maltreating home environment and the strong educational parenting intervention ~PPI!, fo- likelihood that the parents in these families cused on providing mothers with didactic had insecure attachment relationships with training in child development, parenting skills, their own parents ~Crittenden & Ainsworth, coping strategies for managing stresses in the 1989; Gonzalez, Cicchetti, & Rogosch, 2006; immediate environment and assistance in de- Lyons-Ruth & Jacobvitz, 1999; Main & Gold- veloping social support networks. The second wyn, 1984; Main & Hesse, 1990!, maltreated model, infant–parent psychotherapy ~IPP!, in- infants commonly develop disorganized- volved dyadic parent–child therapy sessions disoriented ~Type D! attachments with their designed to improve the parent–child attach- caregivers ~Barnett, Ganiban, & Cicchetti, ment relationship by altering the influence of 1999; Carlson, Cicchetti, Barnett, & Braun- negative maternal representational models on wald, 1989; Lyons-Ruth, Repacholi, McLeod, parent–child interaction. The pre- and post- & Silva, 1991!. Furthermore, investigations intervention functioning of mothers and in- have revealed that in the rare instance of fants in both models of intervention were secure attachment in maltreated infants, compared with the functioning of mothers and these secure attachments are likely to be un- maltreated infants who receive services typi- stable and to eventuate in insecurity ~Cic- cally available through the Department of chetti & Barnett, 1991; Schneider-Rosen et al., Social Services ~DSS!, the community stan- 1985!. dard ~CS! group. A fourth nonmaltreated Organizational theorists posit that the qual- comparison ~NC! group, comprising demo- ity of the resolution of each stage-salient graphically comparable mothers and their non- 626 D. Cicchetti, F. A. Rogosch, and S. L. Toth maltreated infants, also were included to have In a meta-analysis of 12 studies of the ef- a low income standard of comparison for the fectiveness of preventive or therapeutic inter- maltreatment groups. ventions in enhancing parental sensitivity or Home visitation, a component of the two in- children’s attachment security, van IJzendoorn terventions provided herein, has transpired over et al. ~1995! concluded that interventions were the course of the past several decades as an ef- more effective in improving maternal sensitiv- fective model for preventing damage to vul- ity than in fostering children’s attachment se- nerable children ~National Commission to curity. Importantly, van IJzendoorn et al. also Prevent Infant Mortality, 1989; Olds & Kitz- found that the association between parental in- man, 1993; Shirk, Talmi, & Olds, 2000; United ternal working models of attachment figures and States Congress, 1988!. In fact, the United States infant attachment was stronger than the link be- Advisory Board on Child Abuse and Neglect tween maternal sensitivity and attachment and ~1990! identified home-visitation services as the that the former remained significant after sen- best documented strategy for preventing child sitivity was included as a moderator variable. maltreament. Although impressive findings In another meta-analysis of the effectiveness on the effectiveness of home visitation have of preventive interventions for enhancing emerged ~Olds et al., 1997, 1998; Olds, Hen- parental sensitivity and infant attachment se- derson, Chamberlin, & Tatelbaum, 1986!, a re- curity, Bakermans-Kranenburg et al. ~2003! con- cent study on the efficacy of home-visitation cluded that the most effective interventions used services for families where maltreatment had a moderate number of sessions and a behav- already occurred did not succeed in preventing ioral focus ~e.g., van den Boom, 1994, 1995!. the recidivism of physical abuse or neglect However, within the meta-analysis, only three ~MacMillan et al., 2005!. Moreover, to date, randomized studies, all conducted with multi- evaluations of whether home-visitation ser- problem populations, were described as inten- vices can alter the future life-course develop- sive and having numerous sessions. These ment in infants or children who have been interventions also were very comprehensive, maltreated have yet to be conducted. and they included a combination of behavioral, Unlike the PPI model, which focuses on representational, and supportive interventions, current behavior utilizing behavioral interven- thereby making it difficult to disentangle var- tion skills, the core of the IPP intervention ious components of the intervention. resides within the mother’s interactional his- Therefore, the relative efficacy of specific tory and its effect on her representation of types of behavioral versus relationship-based relationships, most significantly, that of her interventions remains open to further investi- infant. A number of attachment-theory in- gation. Accordingly, we compared competing formed interventions have been developed models of intervention, with one being more for high-risk, multiproblem populations ~At- parent-skills oriented ~i.e., PPI! and the other kinson & Goldberg, 2004; Bakermans- being more focused on maternal representa- Kranenburg, van IJzendoorn, & Juffer, 2003, tion and the mother–child relationship ~i.e., 2005; Berlin, Ziv, Amay-Jackson, & Green- IPP!. The evaluation study was guided by the berg, 2005; Cicchetti, Toth, & Rogosch, 1999; following hypotheses: Toth, Maughan, Manly, Spagnola, & Cic- chetti, 2002; van IJzendoorn, Juffer, & Duyvesteyn, 1995!. With respect to attach- 1. Factors associated with higher vulnerabil- ment, theoreticians continue to debate whether ity for insecure infant–mother attachment, modifying parental attachment organization, including maltreatment in the mother’s including their representations of their child, childhood, insecure representations of the will result in improved parenting or, con- mother’s own mother, higher stress, lower versely, whether improving parenting may, in- social support, lower maternal sensitivity, dependent of attention to parental attachment and inappropriate parenting attitudes, will representations, lead to more secure attach- be observed in the mothers in the maltreat- ment relationships between parent and child. ment group. Fostering secure attachment in infants 627

2. At 12 months, the rate of insecure attach- were not targeted for inclusion because of the ment will be higher in the maltreatment limited ongoing contact with their mothers. group than in the nonmaltreatment group. The DHS liaison contacted eligible families 3. Disorganized attachment will predominate and explained the project to mothers. Mothers in the maltreatment group. who were interested in participating signed a 4. Higher rates of change from insecure to release form for their names to be given to secure attachment will be observed in the project staff. During the initial contact by IPP and PPI groups. The rate of secure project staff, the mothers provided informed attachment will increase in the IPP and consent and permission for their infant’s par- PPI groups following the provision of the ticipation. Mothers also signed a release al- interventions. lowing project staff to have access to DHS 5. The rate of secure attachment will be greater records regarding the family’s CPS and pre- in the IPP group than in the PPI group ventive service involvement. following the provision of the interventions. 6. In the absence of active intervention, the Maltreatment determinations. All DHS records rate of stable attachment from 12 to 26 pertaining to the family were accessed and months will be higher in the CS and NC independently coded by trained research cod- groups than in the IPP and PPI groups. ers. The Maltreatment Classification System Stable insecure attachment will predomi- ~MCS; Barnett, Manly, & Cicchetti, 1993! was nate in the CS group. utilized to classify all forms of maltreatment that had occurred in the family and that the infant had experienced. Based on operational Method criteria, the MCS designates all subtypes of maltreatment ~i.e., neglect, emotional maltreat- Participants ment, physical abuse, sexual abuse!. Coding Infants in maltreating families and their moth- of the DSS records was conducted by trained ers were recruited for a study of the effects of research assistants, doctoral students, and clin- child maltreatment on infant development and ical psychologists. Adequate reliability has for an evaluation study of the efficacy of two been obtained ~weighted ks ranging from .86 preventive interventions designed to optimize to .98; Manly, Kim, Rogosch, & Cicchetti, mother–infant relationships and improve par- 2001!. Other investigators have demonstrated enting. The maltreatment sample consisted of that the MCS is reliable and valid in classify- 137 infants ~60 boys, 77 girls! and their moth- ing maltreatment ~Bolger, Patterson, & Kuper- ers. A nonmaltreated comparison group of 52 smidt, 1998; Dubowitz et al., 2005; English infants ~28 boys, 24 girls! and their mothers et al., 2005; Manly, 2005; Smith & Thorn- also were recruited. During the initial assess- berry, 1995!. ment, infants were on average 13.31 months In terms of the subtypes of maltreatment, of age ~SD ϭ 0.81!; mothers ranged in age neglect involves failure to provide for the from 18 to 41 years ~M ϭ 26.87, SD ϭ 5.88!. child’s basic physical needs for adequate food, The majority of mothers ~74.1%! were of mi- clothing, shelter, and medical treatment. In nority race0ethnicity. addition to inadequate attention to physical To recruit 12-month-old infants in maltreat- needs, forms of this subtype include lack of ing families and their mothers, a Department supervision, moral–legal neglect, and educa- of Human Services ~DHS! recruitment liaison tion neglect. Emotional maltreatment involves was retained. As an employee of DHS, the extreme thwarting of children’s basic emo- liaison was able to access DHS Child Protec- tional needs for psychological safety and se- tive Service ~CPS! and preventive service curity, acceptance, and self-esteem, and age- records to identify all infants known to have appropriate autonomy. Examples of emotional been maltreated and0or who were living in maltreatment of increasing severity include be- maltreating families with their biological moth- littling and ridiculing the child, extreme neg- ers. Infants who had been placed in foster care ativity and hostility, exposure to severe marital 628 D. Cicchetti, F. A. Rogosch, and S. L. Toth violence, abandoning the child, and suicidal Following completion of baseline assess- or homicidal threats. Physical abuse involves ments, the mothers and infants recruited for the nonaccidental infliction of physical injury the maltreatment sample were randomly as- on the child ~e.g., bruises, welts, burns, chok- signed to one of three groups: IPP ~n ϭ 53!, ing, broken bones!. Injuries range from minor PPI ~n ϭ 49!,orCS~n ϭ 35!. More cases were and temporary to permanently disfiguring. Fi- randomly assigned initially to the IPP and PPI nally, sexual abuse involves attempted or ac- groups as a result of some mothers declining tual sexual contact between the child and to participate in the interventions. The mal- caregiver for purposes of the caregiver’s sex- treatment groups and the nonmaltreatment ual satisfaction or financial benefit. Events comparison group were comparable on a range range from exposure to pornography or adult of demographic variables ~see Table 1!. The sexual activity, to sexual touching and fon- groups did not differ on child gender, mater- dling, to forced intercourse with the child. nal age, maternal minority race0ethnicity, cur- In the recruited sample, 66.4% of the in- rent receipt of TANF, total family income, or fants had directly experienced abuse and0or marital status. Overall, 53.4% of the infants neglect during the first year of life. None of were girls. Mothers on average were 26.98 the infants with individual level maltreatment years of age ~SD ϭ 5.98!, 74.6% were of mi- had been sexually abused, whereas 8.8% had nority race0ethnicity, and 12.7% of mothers been physically abused, 84.6% had been ne- were currently married. The average family glected, and 69.2% had been emotionally income was $17,151.00, including welfare maltreated. Few cases of physical abuse were benefits, and 96.3% of the families were present in the sample because of the high like- currently receiving TANF. Overall, 41.8% of lihood of foster care placement when abuse the mothers had a high school education or occurs during infancy. The remaining 33.6% less. However, the average Hollingshead of the infants were living in families where level of education differed between groups, abuse and neglect had been experienced by F ~3, 185! ϭ 4.24, p ϭ .001. Post hoc Tukey siblings. tests indicated that mothers in the NC group had higher level education than mothers in the Recruitment of nonmaltreating families. Be- IPP and CS groups. However, the three mal- cause of the predominance of poverty in the treatment groups did not differ from each other. sample of maltreating families, a demograph- Finally, the groups differed on the number of ically comparable sample of mothers and in- children to whom the mothers had given birth, fants in low-income families was recruited F ~3, 188! ϭ 38.82, p , .001. Post hoc tests through the rolls of families receiving Tempo- indicated that mothers in the NC group had rary Assistance to Needy Families ~TANF!. fewer children than the mothers in each of the The DHS liaison identified families with in- maltreatment groups, but the IPP, PPI, and CS fants from the TANF rolls and screened these groups did not differ from each other. Thus, families for child maltreatment. The DHS li- the families in the sample are characterized by aison then contacted eligible families, ex- poverty, relatively large family size, and fre- plained the study, and for interested mothers, quent unstable marriages. obtained a signed release form for the family’s name to be given to the research staff. Sub- Procedures sequently, project staff obtained informed consent and signed permission for all DHS During the baseline assessment period, home- records pertaining to the family to be re- and center-based research sessions were con- viewed. DHS and preventive records were thor- ducted with the mothers and infants. During oughly screened to rule out the existence of interviews conducted in the home by trained documented child maltreatment in these fam- research interviewers, mothers completed de- ilies. Mothers also were interviewed regard- mographic and diagnostic interviews and a ing maltreatment and any DHS involvement variety of self-report questionnaires. During a to further validate nonmaltreatment status. subsequent center-based laboratory session, Fostering secure attachment in infants 629

Table 1. Demographic characteristics

Group

IPP PPI CS NC ~N ϭ 32! ~N ϭ 24! ~N ϭ 81! ~N ϭ 52!

MSD% MSD% MSD% MSD%

Child gender ~% female! 56.6 57.1 54.3 46.2 Maternal Age 27.26 6.52 27.35 6.14 27.77 5.75 25.81 5.39 Minority race0ethnicity 81.1 67.3 77.1 73.1 Education Hollingshead level!*** 3.43 0.80 3.80 0.91 3.43 0.95 4.04 0.86 Receiving TANF 100.0 95.9 97.1 92.3 Total income ~$1000! 17.49 7.59 16.20 8.84 18.25 8.37 16.96 8.15 Marital status Never married 46.9 45.8 49.4 65.4 Married0living with partner 40.6 29.2 33.3 17.3 No longer married 12.5 25.0 17.3 17.3 No. of children*** 3.60 1.76 3.55 1.57 3.60 1.50 1.83 0.92

Note: IPP, infant–parent psychotherapy; PPI, psychoeducational parenting intervention; CS, community standard; NC, nonmaltreated controls. ***p , .001.

mothers and infants participated in the strange evokes affects and memories associated with situation procedure, in addition to other assess- the mother’s childhood relationship experi- ments. Finally, two trained observers, who were ences, and in the process, a mother’s un- unaware of maltreatment status and study hy- resolved and conflictual feelings can be potheses, conducted a 3-hr home observation. projected onto the infant, resulting in dis- Follow-up assessments utilizing the same torted perceptions of the infant, a lack of at- home- and center-based sessions and proce- tunement, and insensitive care. dures were conducted when the children were In IPP, the patient is not the mother or the approximately 26 months old, after the inter- infant, but rather it is the relationship between ventions for the IPP and PPI groups had been the mother and her baby. Masters level thera- completed. pists met weekly with mothers and their 12- month-old infants during sessions conducted in the home over the course of 1 year. The Preventive interventions approach is supportive, nondirective, and non- IPP. This model of intervention is derived didactic, and includes developmental guid- from the work of Fraiberg, Adelson, and Sha- ance based on the mother’s concerns. During piro ~1975!, and it has been shown to be effi- the sessions, the therapist and the mother en- cacious in fostering secure attachment in high gage in joint observation of the infant. The risk, low income, immigrant families ~Lieber- therapist’s empathic responsiveness to the man, 1991, 1992; Lieberman & Pawl, 1988!. mother and the baby allows for expansion of A guiding assumption of IPP is that difficul- parental understanding and exploration of ma- ties in the parent–infant relationship do not ternal misperceptions of the infant. Therapists result from deficits in parenting knowledge strive to allow distorted emotional reactions and skill alone. Rather, the problems that and perceptions of the infant as they are en- maltreating mothers have in relating sensi- acted during mother–infant interaction to be tively and responsively to their infants stem associated with memories and affects from the from insecure internal representational mod- mother’s prior childhood experiences. Through els that evolved in response to the mother’s respect, empathic concern, and unfailing pos- own experiences in childhood. The infant itive regard, the therapeutic relationship pro- 630 D. Cicchetti, F. A. Rogosch, and S. L. Toth vides the mother with a corrective emotional dressed with all mothers, flexibility and lati- experience, through which the mother is able tude on the amount of time spent on each area to differentiate current from past relation- were stressed to tailor the intervention to each ships, form positive internal representations mother’s primary needs. of herself and of herself in relationship to oth- Both the IPP and PPI interventions were ers, particularly her infant. As a result of this manualized, with central components and core process, mothers are able to expand their re- principles of each approach specified. Thera- sponsiveness, sensitivity, and attunement to pists participated in individual and group su- the infant, fostering security in the mother– pervision on a weekly basis, and checks on child relationship and promoting emerging au- the fidelity of the intervention implemen- tonomy in the child. tation for each approach were conducted throughout the course of intervention. Exten- PPI. This model of preventive intervention is sive outreach was typically necessary to en- derived from the preventive intervention work gage mothers in the interventions. The length of Olds and colleagues ~Olds et al., 1997, 1998; of intervention averaged 46.4 weeks ~SD ϭ Olds & Kitzman, 1990!, involving visita- 7.36! for the IPP group and 49.4 weeks ~SD ϭ tion by nurses to the homes of low-income, 4.81! for the PPI group. Although intervention teenage mothers of newborns over a 2-year sessions in the home were scheduled weekly, period. The nurses provided a home-based fewer sessions were conducted as a result of education program on infant physical and psy- cancellations and missed appointments. On av- chological development and parenting, encour- erage, 21.56 ~SD ϭ 9.60! sessions were con- aged mothers to seek further education and ducted in the IPP group and 25.38 sessions employment, and enhanced informal social ~SD ϭ 9.65! in the PPI group. support. The home visitation program was ef- fective in reducing the emergence of child mal- CS. Mothers and infants randomized to the treatment and fostered improved health and CS group did not receive enhanced services. mental health outcomes for mothers and chil- Rather, families continued to receive services dren ~Olds et al., 1997, 1998!. Although very that were typically available to maltreating promising, it is not known whether this ap- families in the community. Families may have proach is effective when maltreatment already continued to be monitored by CPS and may had occurred during infancy. Accordingly, the have participated in other preventive services PPI intervention was supplemented by a vari- programs. ety of cognitive and behavioral techniques to address parenting skill deficits and social– Measures ecological factors, such as limited personal resources, poor social support, and stresses in Demographics interview. Developed by Cic- the home, associated with maltreatment. Mas- chetti and Carlson ~1989!, this measure has ters level therapists, experienced in working been used extensively in research with high- with multiproblem families, conducted home risk, low-income populations. Information ob- visits scheduled weekly over a 12-month pe- tained includes: family composition, gender, riod. The PPI model was psychoeducationally race0ethnicity, birth dates, parent’s education based, striving to address current concerns, and occupation, income level, and receipt of provide parental education and parenting skill public assistance. training, reduce maternal stress, foster social support, and increase life satisfaction. The ap- Childhood Trauma Questionnaire (CTQ). The proach is didactic in nature, providing moth- CTQ ~Bernstein, Fink, Handelsman, & Foote, ers with specific information and knowledge 1994! was designed to obtain retrospective regarding child development. Training in par- self-report information about past experi- enting techniques, problem solving, and relax- ences of maltreatment. Because mothers in ation were utilized. Within a core agenda of the maltreatment groups may have experi- topics on parenting and social skills to be ad- enced maltreatment during their own child- Fostering secure attachment in infants 631 hoods, a measure to address intergenerational attitudes, with particular attention given to mal- transmission of maltreatment was included. adaptive parental behavior associated with Factor analytically derived subscales corre- child abuse and neglect. The items are an- spond to different types of maltreatment, in- swered on a 5-point Likert scale indicating cluding physical0emotional abuse, physical the respondent’s degree of agreement or neglect, emotional neglect, and sexual abuse. disagreement. Four subscales are derived, in- cluding inappropriate expectations, lack of em- Perceptions of Adult Attachment Scale (PAAS). pathy, belief in corporal punishment, and The PAAS ~Lichtenstein & Cassidy, 1991! is a parent–child role reversal. Internal consisten- self-report questionnaire designed to assess cies for the scales in this study ranged from an individuals’ perceptions of the quality of .73 to .90. The AAPI demonstrates discrimi- their early childhood relationships with their nant validity distinguishing between abusive mothers, as well as their current state of mind and nonabusive parents ~Bavolek, 1984!. regarding attachment and their current relation- ships with their mothers. The measure has been Social Support Behaviors Scale (SBS). The shown to have good test–retest reliability and SBS ~Vaux, Riedel, & Stewart, 1987! is a internal consistency ~Lichtenstein & Cassidy, 45-item instrument measuring real and poten- 1991!, and the subscales have demonstrated re- tial social supports available to individuals. lations with the Adult Attachment Interview Five modes are assessed, including emo- ~Lichtenstein-Phelps, Cassidy, Belsky, & Crnic, tional, socializing, practical assistance, finan- 1995!. Subsequent confirmatory factor analy- cial assistance, and advice0guidance. Separate ses have refined the original PAAS subscales scales assess these forms of support for fam- ~Gonzalez et al., 2006!. Subscales character- ily and for friends. Internal consistencies for izing childhood experiences include loved, the scales have exceeded .85. Concurrent va- rejected–neglecting, rejected–threatened, role lidity has been demonstrated through high reversal, and enmeshment. Present state of mind correlations with social support network asso- scales include lack of memory about child- ciations, support appraisals, and the Inventory hood, derogation of attachment, angry, vulner- of Socially Support Behavior. able, and balancing forgiving. Parenting Stress Inventory (PSI). The PSI Maternal Behavior Q-Set (MBQ). After a 3-hr ~Abidin, 1990! is a 101-item questionnaire as- home observation of mother–child interaction, sessing parenting stress in the child domain two trained observers independently com- ~adaptability, acceptability, demandingness, pleted the MBQ ~Pederson & Moran, 1995!. mood, distractibility0hyperactivity, reinforc- The MBQ consists of 90 items that assess ing parent! and the parent domain ~depres- features of maternal sensitivity in relating to sion, attachment, restrictions of role, sense of the infant. Observers sort the 90 items into a competence, social isolation, relationship with forced distribution of nine piles that vary in spouse, parent health!. Internal consistencies the extent to which the item is characteristic range from .70 to .83 for the child domain and or uncharacteristic of maternal behavior. The from .90 to .93 for the parent domain. Internal distribution of items for each mother is corre- consistency for the total stress score is .95. lated with an ideal criterion distribution of Abidin ~1990! reports numerous indicators of maternal sensitivity to derive an individual construct, discriminant, and predictive validity. score. Sensitivity scores evidence stability ~.71! over a 4-month period. Intraclass correlations Strange Situation. During the baseline assess- between pairs of observers averaged .72 at the ments, the standard Strange Situation ~Ains- baseline and .71 at the follow-up assessments. worth et al., 1978! was conducted with mothers and infants to assess the infant’s attachment Adult–Adolescent Parenting Inventory (AAPI). organization. Two independent raters each The AAPI ~Bavolek, 1984! is a 32-item ques- coded all of the videotapes of the individual tionnaire measuring parenting and child-rearing Strange Situation sessions, and raters were un- 632 D. Cicchetti, F. A. Rogosch, and S. L. Toth aware of the maltreatment status and group randomly assigned to the IPP, PPI, or CS assignment of individual mother–child dyads. groups following completion of the baseline When there was a disagreement in coding, assessments. Despite intensive efforts to en- discussion proceeded until consensus was ob- gage mothers assigned to the IPP and the PPI tained for the primary attachment classifica- groups in the interventions, some mothers de- tion. Ainsworth’s criteria for the A, B, and C clined to be involved in the intervention at the classifications were utilized, and D classifica- outset. For other families, mothers could not tions were based on the Main and Solomon be engaged in the intervention. Overall, 21 ~1990! criteria. ~39.6%! of the mothers randomly assigned to During the follow-up postintervention as- the IPP intervention and 25 ~51.0%! of the sessments at age 26 months, the standard mothers assigned to the PPI intervention did Strange Situation was again conducted. The not participate. These rates likely reflect the coding system developed by Schneider-Rosen fact that the families were not seeking treat- et al. ~1985!, which emphasizes the develop- ment. Nevertheless, despite our intentions and mental reorganizations that occur within the efforts, a sizable group of mothers could not attachment behavioral systems as well as be engaged. The rates of refusal are problem- among the attachment, exploration, affilia- atic for intent to treat analyses because the tion, and fear0wariness behavioral systems, lack of participation could overwhelm treat- was utilized. Coding of the videotaped ses- ment effects. sions was completed by two independent rat- To identify potential selection factors that ers who were unaware of group status, family would differentiate those who engaged in the history, or baseline attachment classification. interventions and those who refused, group They coded all videotapes, with any disagree- comparisons were conducted, contrasting the ments resolved by conferencing. The four- IPP group, the IPP decliners, the PPI group, category attachment classification system ~A, the PPI decliners, and the CS group. No sig- B, C, D! was used. nificant group differences among the groups The coders attained 100% agreement with were found for baseline demographic vari- the strange situation training tapes utilized by ables, including child gender, maternal age, Sroufe. One coder ~D.C.! was trained by Mary number of children, marital status, socio- Main to code attachment disorganization0 economic status ~SES!, family income, and disorientation and he trained the second coder race0ethnicity. The groups also did not differ ~F.R.! to reliability. Interrater agreement for significantly on baseline measures of interest, the entire attachment classification coding ex- including maternal representations of the child- ceeded 88%. hood and current relationship with her own mother, history of maltreatment as a child, parenting attitudes, maternal sensitivity, per- Results ceived parenting stress, social support, or his- tory of psychiatric disorders. Thus, evidence Random assignment was not found for selection factors contribut- ing to refusal of the interventions. Because of the preventive goals of reaching Because of the lack of differences between and intervening with the population of maltreat- these maltreatment groups, the primary data ing families with infants, families meeting re- analytic comparisons to be presented involve search criteria were initially identified by the the IPP and PPI cases that engaged in these DHS liaison and invited to participate in the interventions, and a nontreatment group in- research study, with possible involvement in cluding those randomly assigned to the CS the two active interventions. Mothers had not group and those declining the IPP and PPI sought out services on their own, and engage- interventions. These comparisons thus allow ment in the preventive interventions was a for the contrast of intervention effects for those crucial initial step for project therapists. The families who actually completed the interven- families recruited due to maltreatment were tions in relation to maltreating families who Fostering secure attachment in infants 633 did not receive an active intervention. Intent families would differ on historical, represen- to treat analyses were also conducted. In these tational, and contextual constructs, we con- analyses, the original random assignment is ducted a series of multivariate analyses of retained, with the IPP participants and IPP variance ~MANOVAs! with follow-up con- decliners combined, as well as the PPI partici- trasts between the mothers from maltreating pants and PPI decliners combined. families and the nonmaltreating group. The scale scores from the CTQ were first examined to determine whether mothers in Participant retention maltreating families reported higher levels of All families, irrespective of their compliance maltreatment in their own childhoods. A sig- with group assignment, were actively re- nificant multivariate effect for maltreatment cruited for follow-up assessments after the status was found, Wilks’ l ϭ .09, F ~4, 184! ϭ completion of the interventions at age 26 4.77, p ϭ .001. Mothers’ in the maltreatment months. Nevertheless, attrition did occur, and group had significantly higher scores on the 41 families ~21.7%! did not complete post- physical0emotional abuse, emotional abuse, intervention assessments. Although there was physical neglect, and sexual abuse scales, as attrition in each group, the rate of attrition well as the total scale score, than did mothers was higher among families randomly as- in the nonmaltreatment group ~see Table 2!. signed to the CS group ~42.9%!, but did not Mothers representations of their childhood re- differ among the other groups. Families who lationships with their own mothers and their were retained versus lost to follow-up were current state of mind regarding attachment compared on baseline measures to determine also differed significantly, Wilks’ l ϭ .85, variables that might relate to differential attri- F ~10, 155! ϭ 2.82, p ϭ .003. Univariate con- tion. The groups did not differ on major fam- trasts on the PAAS indicated that mothers in ily demographic variables including family the maltreatment group reported that their own SES level, family income, maternal educa- mothers were less loving, more rejecting– tion, maternal age, single marital status, and neglecting, and more rejecting–threatening than minority race0ethnicity. The retained and non- mothers in the nonmaltreatment group. Fur- assessed families also did not differ in terms thermore, mothers in the maltreatment group of baseline measures of maternal representa- reported higher scores for derogation of at- tions of her own mother in childhood and at tachment and for current anger in regard to present, the experience of maltreatment in her their own mothers than did mothers in the own childhood, parenting attitudes, stress as- nonmaltreatment group. The groups did not sociated with parenting, and social support. significantly differ on other PAAS subscales. One variable was identified that did differen- Maternal sensitivity and parenting atti- tiate the groups. Maternal sensitivity was found tudes also distinguished the maltreatment and to be lower in the families not assessed at nonmaltreatment groups, as indicated by a sig- follow-up ~M ϭ 0.21, SD ϭ 0.38! compared to nificant multivariate effect for maltreatment the retained families ~M ϭ 0.35, SD ϭ 0.34!, status, Wilks’ l ϭ .90, F ~5, 178! ϭ 4.14, p ϭ t ~182! ϭ 2.16, p ϭ .03. Aside from this vari- .001. As shown in Table 2, mothers in the able, the sample assessed at follow-up ap- maltreatment group exhibited lower levels of peared representative of the total sample. maternal sensitivity than did mothers in the nonmaltreatment group. Additionally, the scales of the AAPI differed significantly between Baseline functioning of mothers groups, with mothers in the maltreatment group We have already seen that the maltreatment reporting higher inappropriate expectations, groups, including the IPP participants, IPP de- lack of empathy, and acceptance of the appro- cliners, PPI participants, PPI decliners, and priateness of physical punishment than did CS groups did not differ on the variables of mothers in the nonmaltreatment group. Total interest examined in this study. To evaluate AAPI scores also were higher in the maltreat- our hypotheses that the mothers in maltreating ment group. 634 D. Cicchetti, F. A. Rogosch, and S. L. Toth

Table 2. Comparison of maltreatment and nonmaltreatment groups on baseline maternal variables

Nonmaltreated Maltreated ~n ϭ 52! ~n ϭ 137!

MSDMSDt

Childhood Trauma Questionnaire Physical0emotional abuse 2.28 1.02 2.72 1.06 2.59** Emotional neglect 2.23 0.88 2.74 1.02 3.35*** Physical neglect 1.58 0.81 2.11 0.81 4.08*** Sexual abuse 1.91 1.37 2.43 1.50 2.19* Total average score 2.00 0.90 2.50 0.89 3.46*** Perceptions of Adult Attachment Scale Childhood relationship with mother Loved 3.74 0.93 3.17 1.05 3.42*** Rejected-neglected 2.03 1.01 2.74 1.06 4.15*** Rejected-threatened 2.05 0.87 2.50 0.83 3.25*** Role reversal 3.00 0.86 3.16 0.73 1.29 Enmeshed 2.75 0.60 2.77 0.73 0.14 Adult relationship with mother No childhood memories 2.54 1.05 2.77 1.07 1.40 Derogation of attachment 2.13 0.67 2.54 0.74 3.50*** Vulnerable 2.78 0.87 2.93 0.87 1.00 Angry 2.35 1.09 2.87 0.97 3.01** Balancing0forgiving 3.79 0.92 3.60 0.79 1.31 Maternal Behavior Q-Set Maternal sensitivity 0.48 0.29 0.27 0.36 3.68*** Adolescent–Adult Parenting Inventory Inappropriate expectations 1.71 0.61 1.97 0.58 2.63** Lack of empathy 1.94 0.70 2.19 0.75 2.09* Physical punishment 2.04 0.74 2.26 0.69 2.03* Role reversal 2.23 0.95 2.35 0.85 0.83 AAPI total mean score 1.98 0.64 2.19 0.58 2.19* Parenting Stress Inventory Child Stress Scales Distractibility0hyperactivity 2.99 0.53 3.07 0.60 0.90 Reinforces parent ~reversed! 1.48 0.41 1.68 0.46 2.64** Mood 1.89 0.62 2.06 0.70 1.53 Acceptability ~reversed! 1.66 0.50 1.86 0.64 1.99* Adaptability ~reversed! 2.45 0.52 2.67 0.53 2.62** Demandingness 2.04 0.50 2.23 0.56 2.17* Child domain score 2.08 0.34 2.26 0.43 2.64** Parent Stress Scales Sense of competence ~reversed! 2.12 0.43 2.32 0.52 2.44* Attachment ~reversed! 1.74 0.47 1.92 0.52 2.14* Role restriction 2.77 0.76 2.75 0.78 0.14 Depression 2.13 0.70 2.35 0.70 1.89 Relationship with spouse 2.73 0.84 2.71 0.95 0.08 Social isolation 2.19 0.88 2.45 0.76 2.03* Health problems 2.25 0.78 2.58 0.85 2.51* Parent domain score 2.27 0.49 2.44 0.55 1.91 Fostering secure attachment in infants 635

Table 2. ~cont.!

Nonmaltreated Maltreated ~n ϭ 52! ~n ϭ 137!

MSDMSDt

Social Support Behaviors Support from family Emotional 3.95 1.14 3.62 1.28 1.61 Socializing 3.95 1.08 2.51 1.21 2.24* Practical assistance 3.96 1.06 3.47 1.21 2.53* Financial 3.86 1.17 3.33 1.31 2.52* Advice0guidance 4.04 1.06 3.61 1.23 2.19* Total family score 3.95 1.05 3.51 1.19 2.32* Support from friends Emotional 3.87 1.20 3.70 1.17 0.89 Socializing 3.87 1.18 3.67 1.13 1.07 Practical assistance 3.49 1.22 3.31 1.20 0.87 Financial 3.28 1.18 3.10 1.26 0.84 Advice0guidance 3.66 1.18 3.53 1.19 0.62 Total friends score 3.63 1.14 3.46 1.12 0.91

*p , .05. **p , .01. ***p , .001.

Differences in stress associated with par- support from family members in the areas of enting also were found between mothers in socializing, practical assistance, financial as- the maltreatment and nonmaltreatment group. sistance, and advice0guidance. A MANOVAwith PSI subscales as the depen- Thus, mothers in the maltreatment group dent variables revealed a significant multivar- differed across multiple domains associated iate effect for maltreatment status, Wilks’ l ϭ with greater vulnerability for maltreatment and .87, F ~13, 169! ϭ 1.95, p ϭ .03. Univariate problems in parenting and family functioning. contrasts indicated that mothers in the mal- Mothers in the maltreatment group reported treatment group reported more stress in sub- more abuse and neglect in their own child- scales of both the child and the parent domains. hoods, representations of more rejecting, less Mothers in the maltreatment group experi- loving relationships with their own mothers as enced their children as less reinforcing, accept- children, more derogation of attachment gen- ing, and adaptable, and more demanding than erally, and more current anger in their relation- did mothers in the nonmaltreatment group. In ships with their own mothers. Mothers in the terms of the parent domain, mothers in the maltreatment group also were observed to be maltreatment group reported more stress as- less sensitive with their infants and to have a sociated with a low sense of parenting com- range of problematic parenting attitudes. Fi- petence, difficulties in attachment to their nally, greater parenting stress and less social children, social isolation, and health concerns. support among these mothers also were found. In terms of social support, a marginal multi- variate effect for maltreatment status was found Baseline attachment classifications for family support variables, Wilks’ l ϭ .95, F ~5, 181! ϭ 2.09, p ϭ .07, whereas the multi- Across the four experimental groups, the variate effect for support from friends was not distribution of the four-group attachment significant, Wilks’ l ϭ .99, F ~5, 179! ϭ .42, classifications differed significantly, x2 ~9, p ϭ .83. The univariate contrasts for family N ϭ 189! ϭ 63.98, p , .001 ~see Table 3!. support variables were examined, and moth- Follow-up contrasts indicated that no signifi- ers in the maltreatment group reported less cant differences existed among the IPP, PPI, 636 D. Cicchetti, F. A. Rogosch, and S. L. Toth

Table 3. Distribution of preintervention Strange Situation attachment classifications by intervention group

Group

IPP PPI CS NC ~N ϭ 32! ~N ϭ 24! ~N ϭ 81! ~N ϭ 52! Attachment Classification % n % n % n % n

Avoidant 6.3 2 12.5 3 3.7 3 7.7 4 Secure 3.1 1 0.0 0 0.0 0 32.7 17 Resistant 3.1 1 4.2 1 3.7 3 17.3 9 Disorganized 87.5 28 83.3 20 92.6 75 42.3 22

Note: IPP, infant–parent psychotherapy; PPI, psychoeducational parenting intervention; CS, community standard; NC, nonmaltreated controls.

and CS groups in their respective distribu- In these analyses, the CS group was com- tions of attachment classifications. However, promised of mothers and infants who did not each maltreatment group differed signifi- take part in the IPP or PPI interventions. The cantly from the NC group. In the NC group, CS nontreatment group included both cases the distribution of cases reflected the high- randomized to the CS group, as well as cases risk nature of the low-income sample. The who declined participation in the IPP and PPI rate of secure attachment was relatively low interventions. The extent to which these sub- ~32.7%!. Additionally, among infants classi- groups were comparable was evaluated. In this fied as insecure, few cases of Type A ~7.7%! analysis the following groups were contrasted: and Type C ~17.3%! were observed, and 42.3% IPP, PPI, CS as randomized, CS who declined of the NC group was classified as Type D. In IPP, and CS who declined PPI. No differences contrast, across the three maltreatment groups, in baseline attachment distributions were found secure attachment was virtually nonexistent; among these five maltreatment groups, x2 ~12, only one infant in the IPP group was classified N ϭ 137! ϭ 9.04, p ϭ .70. No cases of secure as secure. The vast majority of cases in the attachment were observed in the CS sub- maltreatment groups were classified as Type groups. Furthermore, the rates of disorga- D ~87.5, 83.83, and 92.6%, for the IPP, PPI, nized attachment in the three CS subgroups and CS groups, respectively!. Very low rates were 91.4, 100, and 88.0%, respectively, for of organized insecure classifications ~A and the CS randomized, CS ~IPP declined!, and C! were observed across the maltreatment CS ~PPI declined! subgroups, respectively. groups. Thus, the mother0child dyads randomized to Given the scarcity of secure and organized the CS group, and those declining participa- insecure classifications in the three mal- tion did not differ in their distributions of at- treatment groups, we further contrasted the tachment classification at baseline, and these groups on the rate of disorganized attachment. subgroups did not differ from the IPP and PPI Consistent with the overall distributions, the groups. presence of disorganized attachment differed significantly among the groups, x2 ~3, N ϭ Postintervention attachment classification 189! ϭ 48.57, p , .001. As in the four-group attachment classifications, the rate of disorga- Following the completion of the IPP and PPI nized attachment did not differ among the IPP, preventive interventions, all mother–child dy- PPI, and CS groups, and each maltreatment ads were reassessed in the strange situation group differed significantly from the NC group when children were approximately 26 months ~all ps , .001!. of age. Table 4 provides the distributions of Fostering secure attachment in infants 637

Table 4. Distribution of postintervention Strange Situation attachment classifications by intervention group

Group

IPP PPI CS NC ~N ϭ 28! ~N ϭ 22! ~N ϭ 54! ~N ϭ 44! Attachment Classification % n % n % n % n

Avoidant 7.1 2 0.0 0 18.5 10 13.6 6 Secure 60.7 17 54.5 12 1.9 1 38.6 17 Resistant 0.0 0 0.0 0 1.9 1 4.5 2 Disorganized 32.1 9 45.5 10 77.8 42 43.2 19

Note: IPP, infant–parent psychotherapy; PPI, psychoeducational parenting intervention; CS, community standard; NC, nonmaltreated controls.

attachment classifications for the four treat- stantial increases in secure attachment were ment groups, and again significant group dif- observed in the IPP and PPI groups, whereas ferences were found, x2 ~9, N ϭ148! ϭ 43.75, change in the CS and NC groups was minimal. p , .001. However, the pattern of significant Disorganized attachment continued to be group differences had changed, indicating sub- prominent in the CS group ~77.8%!, contrast- stantial intervention effects. Contrasts among ing with lower rates in the IPP ~32.1%!, PPI the groups revealed that the IPP, PPI, and NC ~45.5%!, and NC ~43.2%! groups, x2 ~3, N ϭ groups all differed significantly ~ ps , .001! 148! ϭ 20.40, p , .001. The intervention and from the CS group, and no significant group NC groups had significantly lower rates of differences existed among the IPP, PPI, and disorganized attachment than the CS group NC groups. The rate of secure attachment in- ~all ps , .01!, and these groups did not differ creased from 3.1 to 60.7% in the IPP group, significantly from each other. The effect sizes and from 0 to 54.5% in the PPI group, whereas for these significant differences were all me- in the CS group, in the absence of theoreti- dium to large ~h ϭ .70–.96!. Compared to cally informed intervention, secure attach- baseline assessments, the rate of disorganized ment at follow-up was virtually nonexistent attachment decreased markedly in the IPP and ~1.9%!. The rate of secure attachment in the PPI groups, whereas little change was ob- NC group was roughly unchanged ~32.7 to served for the CS and NC groups, as shown in 38.6%!. Consistent with the four-group attach- Figure 2. ment classification differences, the rate of se- The pattern of stability and change in at- cure versus insecure attachment among the tachment classification from pre- to postinter- groups was significantly different at postinter- vention also was examined in more detail. Each vention, x2 ~3, N ϭ 148! ϭ 39.35, p , .001, case was classified into one of four groups, with the IPP, PPI, and NC groups having sig- including stable secure, insecure to secure, se- nificantly higher rates of secure attachment than cure to insecure, and stable insecure. When the CS group ~all ps , .001!, and not differing the four experimental groups were contrasted, significantly from one another. The effect size, significant group differences were observed, h, for the contrasts of the CS group with the x2 ~9, N ϭ 148! ϭ 85.28, p , .001, as shown IPP, PPI, and NC groups were 1.51, 1.41, and in Table 5. Follow-up contrasts indicated that 1.17, respectively, indicating that the differ- all groups were significantly different from ences in rates of secure attachment all consti- each other ~all ps , .002!, except for the IPP tuted large effect sizes. Figure 1 depicts the and PPI group contrast. pattern of change for secure versus insecure at- The majority of cases in the IPP and PPI tachment across the four study groups. Sub- groups ~57.1 and 54.5%, respectively! changed 638 .Cchti .A ooc,adS .Toth L. S. and Rogosch, A. F. Cicchetti, D.

Figure 1. The percentage of secure attachment classifications at baseline and postintervention follow-up for the four study groups. otrn eueatcmn ninfants in attachment secure Fostering

Figure 2. The percentage of disorganized–disoriented attachment classifications at baseline and postintervention follow-up for the four study groups. 639 640 D. Cicchetti, F. A. Rogosch, and S. L. Toth

Table 5. Distribution of pre- to postintervention change patterns in Strange Situation attachment classifications by intervention group

Group

IPP PPI CS NC ~N ϭ 28! ~N ϭ 22! ~N ϭ 54! ~N ϭ 44! Pre- to Postintervention Change Pattern % n % n % n % n

Secure to secure 3.6 1 0.0 0 0.0 0 20.5 9 Insecure to secure 57.1 16 54.5 12 1.9 1 18.2 8 Secure to insecure 0.0 0 0.0 0 0.0 0 18.2 8 Insecure to insecure 39.3 11 45.5 10 98.1 53 43.2 19

from insecure to secure attachments. In con- in the IPP ~28.6%!, PPI ~36.4%!, and NC trast, the rate of changing from insecure to ~27.3%! groups, and these groups did not dif- secure attachment was 1.9% in the CS group fer from each other. These significant differ- and 18.9% in the NC group. Overall, the rate ences involved medium to large effect sizes of becoming secure differed significantly ~h ϭ .94, .78, and .87, respectively!. among the groups, x2 ~3, N ϭ 148! ϭ 42.19, p , .001, with all groups differing signifi- Intent to treat analyses cantly ~all ps , .005!, except the IPP and PPI groups. The effect sizes for the contrasts of Because of the frequency of mothers declin- the CS group with the IPP and PPI groups ing to take part in the IPP and PPI interven- were large ~h ϭ 1.51 and 1.46, respectively!. tions, a sizable number of mothers and infants Medium to large effect sizes were found for randomized to the intervention groups did not, the greater likelihood of the IPP ~h ϭ .84! and in fact, complete the interventions. In the in- PPI ~h ϭ .79! groups changing from to inse- tent to treat analyses, we examined group cure to secure, relative to NC group. differences based on the original random as- The experimental groups also differed in signment, thereby including the IPP decliners the rate of stable insecure attachment, x2 ~3, in the IPP group and the PPI decliners in the N ϭ148! ϭ 45.61, p , .001. In the absence of PPI group. Although the groups declining in- intervention, nearly all of the CS group was tervention did not differ on baseline attach- stable insecure ~98.1%!. The rates of stable ment classifications from the mothers0infants insecure attachment were significantly lower randomized to the CS group, whether these in the IPP ~39.3%!, PPI ~45.5%!, and NC subgroups differed on postintervention attach- groups ~43.2%!, all ps , .001, and the rates of ment distributions was initially determined. stable insecure attachment did not differ among At follow-up, the IPP decliners, PPI decliners, these groups. The significant group contrasts and CS randomized groups did not differ on all represented large effect sizes ~h ϭ 1.51, the four-group attachment classifications, x2 1.34, and 1.43, respectively!. ~6, N ϭ 54! ϭ 6.25, p ϭ .40. Disorganized Because of the high rates of disorganized attachment was the most frequent follow-up attachment in this sample, the rate of stable classification for all subgroups, including disorganized attachment also was examined, 85.0% for those declining IPP, 75% for those and significant group differences were found, declining PPI, and 72.2% for those random- x2 ~3, N ϭ 148! ϭ 27.60, p , .001. In the CS ized to CS, and these rates did not differ among group, 74.1% of the children were classified the subgroups, x2 ~2, N ϭ 54! ϭ1.00, p ϭ .61. as disorganized at both time periods. In con- In terms of the pattern of change from base- trast, the rates of stable disorganized attach- line to follow-up, all cases except one across ment were significantly lower ~all ps , .002! the three subgroups were stable insecure. Fur- Fostering secure attachment in infants 641 thermore, all subgroups evinced comparable found, with the IPP and PPI groups having rates of stable D attachment, 80.0, 75.0, and higher rates of secure attachment following 66.7%, respectively, and these rates did not the intervention, higher rates of changing from differ significantly, x2 ~2, N ϭ 54! ϭ 0.89, p ϭ insecure to secure attachment, and lower rates .64. Thus, the three subgroups of cases not of stable disorganized attachment than the CS receiving intervention were comparable and group. However, these differences were driven characterized by stable insecure and disorga- by the cases that actually participated in the nized attachment across the two assessment intervention. periods. Given the lack of change toward secure attachment in the cases declining IPP and PPI, Mediators of intervention efficacy inclusion of these participants in the interven- tion groups to which they were initially ran- We evaluated whether constructs targeted by domized is likely to reduce the large effect the IPP and PPI interventions functioned as sizes observed for intervention effects in pro- mediators of intervention efficacy, explaining moting secure attachment for those mothers the processes through which the interventions and infants who actually participated in the had their effects on promoting attachment se- interventions. When the groups were ana- curity. In this context, mediators of interven- lyzed as randomized in the intent to treat analy- tion efficacy involve variables that changed ses, intervention effects nevertheless continued differentially in the IPP and0or PPI groups to be found, despite the sizable number of and that accounted for the variation in the nonparticipating cases. At follow-up, the rate attachment outcomes that occurred. Thus, the of secure attachment continued to differ across interventions influence change in the mediat- groups, x2 ~3, N ϭ 148! ϭ 12.03, p ϭ .007, ing variable that, in turn, influences the out- with those cases randomized to the IPP come. Processes that were expected to change ~40.9%! and to the PPI ~30.0%! groups and as a result of IPP included improvements in the NC group ~38.8%! having significantly maternal representations of her own mother higher rates of secure attachment than the and increases in maternal sensitivity. For the group randomized to CS ~0%!~contrast ps , PPI intervention, change in parenting atti- .01!. The effect sizes for each of these con- tudes, decreases in child-rearing related stress, trasts with the CS group were large ~h ϭ1.16– and increased social support were targeted 1.39!. Similarly, the rate of changing from areas for change. Thus, initially, whether there insecure to secure attachment at follow-up were changes in these domains from baseline also differed, x2 ~3, N ϭ 148! ϭ 12.66, p ϭ to follow-up and whether there was differen- .005. The randomized IPP ~38.6%! and PPI tial change for the experimental groups was ~30.0%! groups had significantly higher rates evaluated. Repeated measures analysis of vari- of becoming secure than the CS group ~0%! ances with pre- and postintervention assess- ~ ps , .01!, and these contrasts constituted ments of respective variables as the dependent large effect sizes ~h ϭ 1.34 and 1.16, respec- variables were conducted. The primary inter- tively!. Finally, the groups as randomized est was in determining if there were signifi- also differed on the rate of stable disorga- cant Group ϫ Time interactions, which could nized attachment, x2 ~3, N ϭ 148! ϭ 15.79, then be related to the attachment outcomes. In p ϭ .001. Of interest, stable disorganized at- these analyses, none of the variables consid- tachment was more frequent in the group ran- ered, that is, measures of maternal representa- domized to CS ~80.0%! than in the cases tions, maternal sensitivity, parenting attitudes, randomized to IPP ~45.5%! and to PPI ~50.0%; child-rearing stress, social support, was in- p ϭ .01 and .025, h ϭ .83 and .64, respec- volved in the targeted interaction effect. As a tively!. Thus, the intent to treat analyses in- result, none of these constructs could serve as dicate that, even with the cases declining mediator of intervention efficacy, accounting intervention included in the IPP and PPI for how improvements in attachment security groups, treatment effects were nonetheless were attained. 642 D. Cicchetti, F. A. Rogosch, and S. L. Toth

Discussion the maltreatment groups ~,1%! was classi- fied as securely attached. Although the rate of Consistent with our hypothesis, mothers in the secure attachment was significantly higher in maltreatment groups compared to those in the the group of infants in nonmaltreating fami- nonmaltreatment group were found to differ lies ~32.7%!, this was well below the per- substantially on important constructs expected centage of secure attachment observed in to confer vulnerability on the mother’s capac- nondisadvantaged samples of babies ~Thomp- ity to form a secure attachment relationship son, 1998!. The low rate of attachment secu- with her infant. Specifically, mothers in the rity found in the group of nonmaltreated maltreatment groups reported having experi- comparison infants underscores the extremely enced more abuse and neglect in their own high-risk backgrounds in which these babies childhoods; moreover, mothers in the mal- reside. Clearly, the stresses associated with treatment groups had more negative represen- poverty and other risk factors that commonly tations of aspects of their childhood and accompany membership in the low socioeco- contemporary relationships with their own nomic strata increase the probability that strains mothers. will be placed on the mother–infant attach- It is not surprising that, given their early re- ment relationship. lationship difficulties and current anger and re- In terms of the specific insecure attach- sentment to their own mothers, maltreating ment organizations observed, not only were mothers also reported less availability of so- the infants in maltreating families classified cial support from family members. Accord- as insecure, but also 89.8% were classified ingly, the ability of these mothers to rely on as disorganized. In contrast, 42.3% of the in- family members in times of need appears com- fants in nonmaltreating families were clas- promised. Maltreatment group mothers also re- sified as disorganized. Thus, in the baseline ported significantly higher current stress with assessments of the largest group of infants in respect to feeling more demands and struggles maltreating families assessed in the Strange in relation to their infant, as well as feeling less Situation to date ~N ϭ 137!, disorganized at- competent as a parent. Social isolation and health tachment organization was almost ubiquitons. concerns also were more prominent stressors The high rate of disorganized attachment found for these mothers. The mothers in the maltreat- in the infants in maltreating families is consis- ment group also conveyed deficits in their un- tent with the percentage of Type D attachment derstanding of appropriate parenting attitudes in maltreated infants reported in the extant and behavior relative to nonmaltreating moth- literature ~Bakermans-Kranenburg et al., 2003; ers. Moreover, based on extensive home obser- Barnett et al.,1999; Carlson et al., 1989; Lyons- vations of mother–infant interaction, research Ruth et al., 1991!. No differences in the rate assistants who were unaware of group status of insecure or disorganized attachment were and experimental hypotheses rated mothers in observed among the three groups of babies the maltreatment group as substantially lower from maltreating families. In addition, the rate in maternal sensitivity to their infants than non- of disorganized attachment in the group of maltreating mothers. Thus, childhood histories infants from nonmaltreating families was of abuse and neglect, negative relationship rep- higher than usually found in infants from the resentations, limited family social support, low SES ~Barnett et al., 1999; Carlson et al., stressors in multiple domains, and insensitive 1989!, attesting to the high-risk status of the maternal patterns of relating to her infant likely comparison group. Furthermore, the high rate conspire to impair the development of secure of disorganized attachment in infants from attachment relationship formation with their maltreating families places them at extremely infant. high risk for embarking on negative develop- The baseline attachment classifications in- mental trajectories; especially given the dele- dicated that the babies in maltreating families terious sequelae of disorganized attachment exhibited an extremely high rate of insecure ~Carlson, 1998; Green & Goldwyn, 2002; van attachment. Specifically, only one infant in IJzendoorn et al., 1999!. Fostering secure attachment in infants 643

At the conclusion of the intervention, when individually administered to child partici- the infants were approximately 26 months of pants by experimenters unaware of group age, dramatic changes in attachment classifi- status and experimental hypotheses. The nar- cation were observed. At intervention follow- ratives were videotaped and subsequently up, the rate of secure attachment had increased coded. The narratives depicted moral dilem- remarkably in the two intervention groups to mas and emotionally charged events in the 60.7 and 54.5% for the IPP and PPI groups, context of parent–child and family relation- respectively. In contrast, the rate of secure ships. Narrative story stems included vig- attachment in the CS group remained virtu- nettes designed to elicit children’s perceptions ally nonexistent ~1.9%!. Moreover, the rate of of the parent–child relationship, of self, and secure attachment in the NC group ~38.6%! of maternal behavior in response to child trans- continued to surpass the CS group. Thus, gressions, intrafamilial conflicts, and child marked gains in establishing secure attach- accidents. ment organization were achieved in both of Children in the PPP intervention evidenced the intervention groups. a greater decline in maladaptive maternal rep- The results of this randomized preven- resentations over time than did children in the tive intervention trial demonstrate that an PPI and CS interventions. Moreover, children attachment-theory informed intervention, IPP, who took part in the PPP intervention dis- and an intervention that focuses on improving played a greater decrease in negative self- parenting skills, increasing maternal knowl- representations than did children in the CS, edge of child development, and enhancing the PPI, and NC groups. Additionally, the mother– coping and social support of skills of maltreat- child relationship expectations of PPP chil- ing mothers, PPI, both were successful in al- dren became more positive over the course of tering the predominantly insecure attachment the intervention compared with children in the organizations of infants in maltreating fami- PPI and NC groups. These results suggest that lies. These findings are contrary to our hypoth- an attachment-theory informed model of inter- esis that the IPP intervention would be more vention ~PPP! is more effective at improving successful in improving attachment security representations of self and of caregivers than would the PPI intervention. Perhaps than is a didactic model of intervention ~PPI! differences in the future outcomes of these directed at parenting skills and contradict pre- interventions will be discovered during the dictions that would emanate from the meta- preschool period as development and attach- analysis of interventions targeting maternal ment organization become increasingly repre- sensitivity and child attachment ~Bakermans- sentational in nature. Kranenberg et al., 2003, 2005!. Because the In support of this assertion, Toth and col- intervention focused on changing representa- leagues ~2002! conducted a preventive inter- tional models, as assessed by a narrative story- vention trial with maltreating mothers and their stem measure, outcomes that might be expected preschool age maltreated children. Maltreated to improve more dramatically in the PPI model children were randomly assigned to an attach- ~e.g., parenting skills, knowledge of child de- ment theory-informed intervention, PPP, PPI, velopment! could not be addressed. Alterna- or the CS. An NC group of infants also par- tively, the PPP intervention may have been ticipated. Both the PPP and PPI interventions more effective than the PPI intervention as a were modified to address the advances in sym- result of being implemented during the pre- bolic and representational development that school period when representational capaci- occur during the preschool period. ties are expanding. In contrast, the IPP and At baseline ~M age ϭ 48.18 months! and PPI interventions may have both resulted again at postintervention 1 year later ~M age ϭ in improvements in attachment security be- 61.47 months!, 11 narrative story stems cause they were initiated during infancy and ~Bretherton, Oppenheim, Buchsbaum, Emde, toddlerhood. & The MacArthur Narrative Group, 1990; In the current study, change from insecure Bretherton, Ridgeway, & Cassidy, 1990! were to secure attachment was significantly more 644 D. Cicchetti, F. A. Rogosch, and S. L. Toth likely in the IPP and PPI groups than in the CS competent resolutions of subsequent devel- and the NC groups. In contrast to the reorga- opmental tasks. In the context of a secure at- nization in attachment that had occurred in the tachment relationship, the secure maltreated intervention groups, stability of insecure at- children are more likely to develop positive tachment was almost universal in the CS group self-representations. Secure representational ~98.1%!. Similarly, stability of attachment also models of the self and self in relation to others was more common in the NC group with 70.4% then will further promote competent striv- of the insecure nonmaltreated children remain- ing to adapt to subsequent developmental ing insecure at postintervention follow-up and challenges as these children begin forming 52.9% of those who were secure remaining relationships with other adults and peers secure. Overall, stability of secure0insecure ~Sroufe & Fleeson, 1986, 1988!. By interven- attachment was 63.6% in the NC group. As ing to promote more competent developmen- such, continuity of attachment organization was tal trajectories through instilling a secure more characteristic of the groups not partici- attachment organization, it is anticipated that pating in the active preventive interventions. maladjustment and the development of psy- Given the extremely high rate of disorga- chopathology will be more likely to be averted nized attachment in the maltreated infants at as these children develop. age 13 months, it also is remarkable that the Given the success of this randomized pre- two preventive interventions were efficacious vention trial, it becomes important to consider in reducing this atypical attachment pattern. why these therapeutic models were effective Stable disorganized attachment was observed in altering attachment security when previous among 74.1% of the CS children, whereas sta- investigations did not yield such findings. A ble disorganized attachment occurred at much number of components of these interventions lower rates in the two intervention groups, may have contributed to their success. All 28.6 and 36.4% for the IPP and PPI groups, therapists received extensive training before respectively, comparable to the rate observed implementing the interventions and these cli- in the NC group ~27.3%!. Thus, infants who nicians were familiar not only with the inter- have been maltreated are highly likely to main- vention modality, but also with the theory from tain disorganized0disoriented attachments in which the interventions was derived. All ther- the absence of intensive efforts to improve the apists also had considerable prior experience mother–child relationship and parenting. working with low-income maltreating fami- Taken together, the results of the interven- lies. Both models also were manualized, tion provide strong support for the benefits of weekly individual and group supervision was the preventive interventions in altering the de- provided, and therapist adherence to their re- velopmental trajectories of maltreated infants. spective model was monitored for each case Through targeting a central developmental task throughout the provision of the intervention. of the infancy period ~Sroufe, 1979!, the inter- Case loads also were maintained at levels con- ventions were successful in transforming the siderably lower than typical of outpatient men- attachment organization of a substantial per- tal health settings, and therapists were therefore centage of maltreated infants. Not only were able to devote considerable time to engaging marked reductions achieved in the rate of in- mothers and to conceptualizing treatment plans. secure attachment, but also disorganized at- The positive outcome of this investigation sup- tachments were shown to be modifiable and ports the importance of investing in more costly secure attachments were attained. These re- interventions, including allowing therapists suf- sults are noteworthy for demonstrating the mal- ficient time for training and supervision. The leability and plasticity of the attachment system extensive outreach expended toward engag- through focusing on changing aspects of the ing mothers in treatment also is a noteworthy early mother–child relationship. The establish- component of these interventions that may have ment of secure attachment relationships in the contributed to their success. When mothers maltreated youngsters through the preventive cancelled or failed to be available for sched- interventions holds promise for achieving more uled visits, sessions were rescheduled during Fostering secure attachment in infants 645 the same week. Thus, unlike clinics where who completed the interventions are partially cases are closed when sessions are repeatedly self-selected. Intent to treat analyses examined failed, our therapists continued to reach out to group composition based on the initial random engage mothers in the treatment process. assignment irrespective of whether mothers ac- Clearly, consistent with the principles of tually participated. Consequently, in our sta- the organizational perspective, and with Gott- tistical analyses, those who completed and those lieb’s ~1992! notion of probabilistic epigene- who declined or discontinued their assignment sis, the early insecure, generally disorganized to one of the active preventive interventions were attachments displayed by maltreated infants, combined to examine group differences for those do not doom these youngsters to have poor- we intended to treat. The incorporation of in- quality relationship expectations and negative tent to treat analyses did not alter the findings self-representations throughout the life span reported above; infants in both active preven- ~Cicchetti & Tucker, 1994!. The success of tive interventions were significantly more likely the interventions, informed by basic research to change from insecure to secure attachment. knowledge on the etiology and developmental Moreover, the reductions in disorganized at- sequelae of child maltreatment, suggests that tachment in the active intervention groups attachment organization is modifiable, even if likewise remained even when intent to treat a high percentage of Type D attachments is analyses were conducted. initially characteristic of the sample. Follow- Despite the positive results obtained with ing the organizational perspective, it is ex- the IPP and PPI interventions, there are limi- pected that these maltreated youngsters, now tations to the study. Changes were not found that they are traversing a more positive devel- for the constructs hypothesized to serve as opmental trajectory, will be more likely to con- potential mediators of intervention efficacy. tinue on an adaptive pathway and successfully This state of affairs is common in randomized resolve future salient developmental tasks. The clinical trial research, although intervention preventive interventions have demonstrated efficacy is demonstrated, the mechanisms of that behavioral plasticity is possible, at least change are difficult to identify. In contrast to in the early years of life. the findings of van IJzendoorn et al. ~1995!, The extremely high stability of attachment the current study was successful in improving insecurity in the CS group is especially alarm- young children’s attachment organization, but ing. These results portend that the maltreating enhancement of observed maternal sensitivity infants in the CS group are at exceedingly was not demonstrated. We plan to probe fur- high risk for maladaptive outcomes on sub- ther for changes in the intervention groups sequent stage-salient issues of development. that may explain how the efficacy of the inter- The typical intervention services provided to vention was achieved. Videotaped interactions maltreated infants are not sufficient to reme- between mothers and infants in stressful and diate their early developmental difficulties. unstressful situations will be examined for a Without the receipt of theoretically derived, more detailed analysis of changes in maternal intensive interventions, maltreated infants are behavior and relationship qualities that may likely to embark on a negative developmental account for the improvements in attachment trajectory. Clearly, case management, which security that were observed. focuses upon monitoring the physical safety Another study limitation involves the de- of maltreated children, is inadequate to foster gree to which mothers did not engage in the positive socioemotional development. More- interventions, despite active outreach efforts, over, these results underscore the criticality of indicating that additional strategies are neces- providing evidence-based services to maltreat- sary to facilitate active participation. It is im- ing families. portant to emphasize that all mothers were In our clinical prevention trial, cases ran- solicited for involvement in the interventions domly assigned to IPP and PPI did not always without seeking assistance on their own, and comply with the assignment, declining to be internal motivation for change and improved involved in the interventions.As a result, those family function may not have been present. In 646 D. Cicchetti, F. A. Rogosch, and S. L. Toth essence, we conducted a population study in secure attachment, costlier interventions such which all eligible families were approached as foster care placement, special education ser- rather than only those referred for, mandated vices, residential treatment, and incarceration to, or voluntarily seeking assistance. In view can be averted. Unfortunately, our results also of this, the positive outcomes of these inter- shed light on the harsh reality of the ineffec- ventions are even more remarkable. tiveness of services currently being provided In closing, the translation of knowledge in many communities throughout the nation. from the field of developmental psychopathol- It is critical that professionals, government ogy into the conduct of this clinical trial un- officials, social policy advocates, and mental derscores the importance of broadening such health insurers recognize the necessity of efforts. We believe that the results of this ran- investing in the delivery of theoretically in- domized prevention trial are both gratifying formed, evidence-based interventions. Al- and sobering. The fact that plasticity is possi- though this may appear to be a costly endeavor ble during infancy and that even the most dis- in the short term, the long-term benefits with organized form of attachment is modifiable in respect to enhancing positive child develop- extremely dysfunctional mother–child dyads ment, preventing the dissolution of families, offers significant hope for thousands of young and decreasing the burden of mental illness in children and for their families. By fostering society cannot be overstated.

References

Abidin, R. ~1990!. Parenting stress inventory manual. spective measure of child abuse and neglect. Ameri- Odessa, FL: Psychological Assessment Resources. can Journal of Psychiatry, 151, 1132–1136. Ainsworth, M. D. S., Blehar, M. C., Waters, E., & Wall, Bolger, K. E., Patterson, C. J., & Kupersmidt, J. B. ~1998!. S. ~1978!. Patterns of attachment: A psychological Peer relationships and self-esteem among children study of the Strange Situation. Hillsdale, NJ: Erlbaum. who have been maltreated. Child Development, 69, Atkinson, L., & Goldberg, S. ~Eds.!. ~2004!. Clinical 1171–1197. applications of attachment. Mahwah, NJ: Erlbaum. Bretherton, I., Oppenheim, D., Buchsbaum, H., Emde, R. N., Azar, S. T. ~2002!. Parenting and child maltreatment. In & The MacArthur Narrative Group. ~1990!. Mac- M. H. Bornstein ~Ed.!, Handbook of parenting: So- Arthur Story Stem Battery. Unpublished manuscript. cial conditions and applied parenting ~2nd ed., Vol. 4, Bretherton, I., Ridgeway, D., & Cassidy, J. ~1990!. As- pp. 361–388!. Mahwah, NJ: Erlbaum. sessing internal working models of the attachment Bakermans-Kranenburg, M. J., van IJzendoorn, M. H., & relationship: An attachment story completion task for Juffer, F. ~2003!. Less is more: Meta-analysis of sen- 3-year-olds. In M. Greenberg, D. Cicchetti & E. M. sitivity and attachment interventions in early child- Cummings ~Eds.!, Attachment in the preschool years hood. Psychological Bulletin, 129, 195–215. ~pp. 273–308!. Chicago: University of Chicago Press. Bakermans-Kranenburg, M. J., van IJzendoorn, M. H., & Carlson, E. A. ~1998!. A prospective longitudinal study of Juffer, F. ~2005!. Disorganized infant attachment and attachment disorganization0disorientation. Child De- preventive interventions: A review and meta-analysis. velopment, 69, 1107–1128. Infant Mental Health Journal, 26, 191–216. Carlson, V., Cicchetti, D., Barnett, D., & Braunwald, K. Barnett, D., Ganiban, J., & Cicchetti, D. ~1999!. Maltreat- ~1989!. Disorganized0disoriented attachment relation- ment, negative expressivity, and the development of ships in maltreated infants. Developmental Psychol- Type D attachments from 12- to 24-months of age. ogy, 25, 525–531. Monographs of the Society for Research in Child De- Cerezo, M. A. ~1997!. Abusive family interaction: A re- velopment, 64, 97–118. view. Aggression and Violent Behavior, 2, 215–240. Barnett, D., Manly, J. T., & Cicchetti, D. ~1993!. Defining Cicchetti, D. ~1993!. Developmental psychopathology: child maltreatment: The interface between policy and Reactions, reflections, projections. Developmental Re- research. In D. Cicchetti & S. L. Toth ~Eds.!, Child view, 13, 471–502. abuse, child development, and social policy ~pp. 7– Cicchetti, D. ~2002!. The impact of social experience on 73!. Norwood, NJ: Ablex. neurobiological systems: Illustration from a construc- Bavolek, S. J. ~Ed.!. ~1984!. Handbook for the Adult– tivist view of child maltreatment. Cognitive Develop- Adolescent Parenting Inventory (AAPI). Park City, ment, 17, 1407–1428. UT: Family Developmental Resources, Inc. Cicchetti, D., & Barnett, D. ~1991!. Attachment organiza- Berlin, L. J., Ziv, Y., Amay-Jackson, L., & Greenberg, tion in pre-school-aged maltreated children. Develop- M. T. ~Eds.!. ~2005!. Enhancing early attachments: ment and Psychopathology, 3, 397–411. Theory, research, intervention, and policy. New York: Cicchetti, D., & Carlson, V. ~Eds.!. ~1989!. Child mal- Guilford Press. treatment: Theory and research on the causes and Bernstein, D. P., Fink, L., Handelsman, L., & Foote, J. consequences of child abuse and neglect. New York: ~1994!. Initial reliability and validity of a new retro- Cambridge University Press. Fostering secure attachment in infants 647

Cicchetti, D., & Hinshaw, S. P. ~Eds.!. ~2002!. Prevention abuse and neglect. New York: Cambridge University and intervention science: Contributions to develop- Press. mental theory. Development and Psychopathology, 14, Cummings, E. M., & Davies, P. T. ~1996!. Emotional 667–981. security as a regulatory process in normal develop- Cicchetti, D., & Lynch, M. ~1993!. Toward an ecological0 ment and the development of psychopathology. De- transactional model of community violence and child velopment and Psychopathology, 8, 123–139. maltreatment: Consequences for children’s develop- DeBellis, M. D. ~2001!. Developmental traumatology: ment. Psychiatry, 56, 96–118. The psychobiological development of maltreated Cicchetti, D., & Manly, J. T. ~Eds.!. ~2001!. Operational- children and its implications for research, treatment, izing child maltreatment: Developmental processes and policy. Development and Psychopathology, 13, and outcomes. Development and Psychopathology, 13, 539–564. 755–1048. Dubowitz, H., Pitts, S. C., Litrownik, A. J., Cox, C. E., Cicchetti, D., & Rogosch, F. A. ~1999!. Conceptual and Runyan, D., & Black, M. M. ~2005!. Defining child methodological issues in developmental psychopathol- neglect based on child protective services data. Child ogy research. In P. C. Kendall, J. N. Butcher, & G. N. Abuse and Neglect, 29, 493–511. Holmbeck ~Eds.!, Handbook of research methods in Egeland, B., & Sroufe, L. A. ~1981!. Developmental se- clinical psychology ~pp. 433–465!. New York: Wiley. quelae of maltreatment in infancy. New Directions for Cicchetti, D., & Schneider-Rosen, K. ~1986!. An orga- Child Development, 11, 77–92. nizational approach to childhood depression. In M. English, D. J., Upadhyaya, M. P., Litrownik, A. J., Mar- Rutter, C. Izard, & P. Read ~Eds.!, Depression in shall, J. M., Runyan, D. K., Graham, J. C., et al. ~2005!. young people, clinical and developmental perspec- Maltreatment’s wake: The relationship of maltreat- tives ~pp. 71–134!. New York: Guilford Press. ment dimensions to child outcomes. Child Abuse and Cicchetti, D., & Sroufe, L. A. ~1978!. An organizational Neglect, 29, 597–619. view of affect: Illustration from the study of Down’s Fraiberg, S., Adelson, E., & Shapiro, V. ~1975!. Ghosts in syndrome infants. In M. Lewis & L. Rosenblum ~Eds.!, the nursery: A psychoanalytic approach to impaired The development of affect ~pp. 309–350!. New York: infant–mother relationships. Journal of the American Plenum Press. Academy of Child Psychiatry, 14, 387–421. Cicchetti, D., & Toth, S. L. ~1992!. The role of develop- Gottlieb, G. ~1992!. Individual development and evolu- mental theory in prevention and intervention. Devel- tion: The genesis of novel behavior. New York: Ox- opment and Psychopathology, 4, 489–493. ford University Press. Cicchetti, D., & Toth, S. L. ~in press!. A developmental Gonzalez, A., Cicchetti, D., & Rogosch, F. A. ~2006!. psychopathology perspective on preventive interven- Mental representations of attachment and their rela- tions with high risk children and families. In A. Ren- tion to the intergenerational transmission of child mal- ninger & I. Sigel ~Eds.!, Handbook of child psychology treatment. Unpublished manuscript. ~6th ed.!. New York: Wiley. Green, J., & Goldwyn, R. ~2002!. Attachment disorgani- Cicchetti, D., Toth, S. L., & Bush, M. ~1988!. Develop- zation and psychopathology: New findings in attach- mental psychopathology and incompetence in child- ment research and their potential implications for hood: Suggestions for intervention. In B. Lahey & A. developmental psychopathology in childhood. Jour- Kazdin ~Eds.!, Advances in clinical child psychology nal of Child Psychology and Psychiatry, 43, 835–845. ~pp. 1–71!. New York: Plenum Press. Gunnar, M. R., & Vazquez, D. M. ~2006!. Stress neuro- Cicchetti, D., Toth, S. L., & Rogosch, F. A. ~1999!. The biology and developmental psychopathology. In D. efficacy of toddler–parent psychotherapy to increase Cicchetti & D. Cohen ~Eds.!, Developmental psycho- attachment security in offspring of depressed moth- pathology: Developmental neuroscience ~2nd ed., ers. Attachment and Human Development, 1, 34–66. Vol. 2!. New York: Wiley. Cicchetti, D., & Tucker, D. ~1994!. Development and Hinshaw, S. P. ~2002!. Intervention research, theoretical self-regulatory structures of the mind. Development mechanisms, and causal processes related to external- and Psychopathology, 6, 533–549. izing behavior problems. Development and Psychopa- Cicchetti, D., & Valentino, K. ~2006!. An ecological trans- thology, 14, 789–818. actional perspective on child maltreatment: Failure of Hofer, M. A. ~1987!. Early social relationships: A the average expectable environment and its influence psychobiologist’s view. Child Development, 58, upon child development. In D. Cicchetti & D. J. Co- 633–647. hen ~Eds.!, Developmental psychopathology. Risk, dis- Howe, G. W., Reiss, D., & Yuh, J. ~2002!. Can prevention order, and adaptation ~2nd ed., Vol. 3!. New York: trials test theories of etiology? Development and Psy- Wiley. chopathology, 14, 673–694. Coie, J. D., Watt, N. F., West, S. G., Hawkins, D., Asar- Howes, P. W., & Cicchetti, D. ~1993!. A family0relational now, J. R., Markman, H. J., et al. ~1993!. The science perspective on maltreating families: Parallel pro- of prevention: A conceptual framework and some di- cesses across systems and social policy implications. rections for a National Research Program. American In D. Cicchetti & S. L. Toth ~Eds.!, Child abuse, child Psychologist, 48, 1013–1022. development, and social policy ~pp. 249–300!. Nor- Crittenden, P. M. ~1988!. Relationships at risk. In J. Bel- wood, NJ: Ablex. sky & T. Nezworski ~Eds.!, Clinical implications of Ialongo, N., Rogosch, F. A., Cicchetti, D., Toth, S. L., attachment theory ~pp. 136–174!. Hillsdale, NJ: Buckley, J., Petras, H., et al. ~2006!. The importance Erlbaum. of utilizing and translating basic research knowledge Crittenden, P. M., & Ainsworth, M. D. S. ~1989!. Child on normal and abnormal development into preventive maltreatment and attachment theory. In D. Cicchetti interventions. In D. Cicchetti & D. Cohen ~Eds.!, & V. Carlson ~Eds.!, Child maltreatment: Theory and Developmental psychopathology: Theory and method research on the causes and consequences of child ~2nd ed., Vol. 1!. New York: Wiley. 648 D. Cicchetti, F. A. Rogosch, and S. L. Toth

Institute of Medicine. ~1994!. Reducing risks for mental Main, M., & Solomon, J. ~1990!. Procedures for identify- disorders: Frontiers for preventive intervention re- ing infants as disorganized0disoriented during the search. Washington, DC: National Academy Press. Ainsworth Strange Situation. In M. Greenberg, D. Kellam, S. G., & Rebok, G. W. ~1992!. Building devel- Cicchetti & E. M. Cummings ~Eds.!, Attachment in opmental and etiological theory through epidemiolog- the preschool years ~pp. 121–160!. Chicago: Univer- ically based preventive intervention trials. In J. McCord sity of Chicago Press. & R. E. Tremblay ~Eds.!, Preventing antisocial be- Manly, J. T. ~2005!. Advances in research definitions of havior: Interventions from birth through adolescence child maltreatment. Child Abuse and Neglect, 29, ~pp. 162–195!. New York: Guilford Press. 425–439. Kraemer, H. C., Kazdin, A. E., Offord, D. R., Kessler, Manly, J. T., Kim, J. E., Rogosch, F. A., & Cicchetti, R. C., Jensen, P. S., & Kupfer, D. J. ~1997!. Coming to D. ~2001!. Dimensions of child maltreatment and terms with the terms of risk. Archives of General children’s adjustment: Contributions of developmen- Psychiatry, 54, 337–343. tal timing and subtype. Development and Psychopa- Lichtenstein, J., & Cassidy, J. ~1991, March!. The Inven- thology, 13, 759–782. tory of Adult Attachment: Validation of a new mea- National Commission to Prevent Infant Mortality. ~1989!. sure. Paper presented at the Society for Research in Home visiting: Opening doors for America’s pregnant Child Development, Seattle, WA. women and children. Washington, DC: Author. Lichtenstein-Phelps, J., Cassidy, J., Belsky, J., & Crnic, K. Noam, G. ~1992!. Development as the aim of clinical ~1995!. Introduction of a new measure of adult attach- intervention. Development and Psychopathology, 4, ment. Unpublished manuscript. 679–696. Lieberman, A. F. ~1991!. Attachment theory and infant– Olds, D., Eckenrode, J., Henderson, C., Kitzman, H., Pow- parent psychotherapy: Some conceptual, clinical, and ers, J., Cole, R., et al. ~1997!. Long-term effects of research considerations. In D. Cicchetti & S. L. Toth home visitation on maternal life course and child abuse ~Eds.!, Rochester Symposium on Developmental and neglect: Fifteen-year follow-up of a randomized Psychopathology: Models and integrations ~Vol. 3, trial. Journal of the American Medical Association, pp. 261–287!. Rochester, NY: University of Roch- 278, 637–643. ester Press. Olds, D., Henderson, C., Kitzman, H., Eckenrode, J., Cole, Lieberman, A. F. ~1992!. Infant–parent psychotherapy R., & Tatelbaum, R. ~1998!. The promise of home with toddlers. Development and Psychopathology, 4, visitation: Results of two randomized trials. Journal 559–574. of Community Psychology, 26, 5–21. Lieberman, A. F., & Pawl, J. H. ~1988!. Clinical appli- Olds, D., Henderson, C. R., Chamberlin, R., & Tatel- cations of attachment theory. In J. Belsky & T. Nez- baum, R. ~1986!. Preventing child abuse and neglect: worski ~Eds.!, Clinical implications of attachment A randomized trial of nurse home visitation. Pediat- ~pp. 325–351!. Hillsdale, NJ: Erlbaum. rics, 78, 65–78. Lyons-Ruth, K., Easterbrooks, M. A., & Cibelli, C. D. Olds, D. L., & Kitzman, H. ~1990!. Can home visitation ~1997!. Infant attachment strategies, infant mental lag, improve the health of women and children at environ- and maternal depressive symptoms: Predictors of in- mental risk? Pediatrics, 86, 108–116. ternalizing and externalizing problems at age 7. De- Olds, D. L., & Kitzman, H. ~1993!. Review of research on velopmental Psychology, 33, 681–692. home visiting for pregnant women and parents of Lyons-Ruth, K., & Jacobvitz, D. ~1999!. Attachment dis- young children. Future of Children, 3, 53–92. organization: Unresolved loss, relational violence, and Pederson, D. R., & Moran, G. ~1995!. A categorical de- lapses in behavioral and attentional strategies. In J. scription of infant–mother relationships in the home Cassidy & P. R. Shaver ~Eds.!, Handbook of attach- and its relation to Q-sort measures of infant–mother ment: Theory, research, and clinical applications interaction. Monographs of the Society for Research ~pp. 520–554!. New York: Guilford Press. in Child Development, 60, 111–132. Lyons-Ruth, K., Repacholi, B., McLeod, S., & Silva, E. Pipp, S., & Harmon, R. J. ~1987!. Attachment as reg- ~1991!. Disorganized attachment behavior in infancy: ulation: A commentary. Child Development, 58, Short-term stability, maternal and infant correlates, 648–652. and risk-related subtypes. Development and Psycho- Pollak, S. D., Cicchetti, D., Klorman, R., & Brumaghim, pathology, 3, 377–396. J. ~1997!. Cognitive brain event-related potentials and MacMillan, H. L., Thomas, B. H., Jamieson, E., Walsh, emotion processing in maltreated children. Child De- C. A., Boyle, M. H., Shannon, H. S., et al. ~2005!. velopment, 68, 773–787. Effectiveness of home visitation by public-health Rogosch, F. A., Cicchetti, D., Shields, A., & Toth, S. L. nurses in prevention of the recurrence of child physi- ~1995!. Parenting dysfunction in child maltreatment. cal abuse and neglect: A randomized controlled trial. In M. H. Bornstein ~Ed.!, Handbook of parenting ~Vol. Lancet, 365, 1786–1793. 4, pp. 127–159!. Hillsdale, NJ: Erlbaum. Main, M., & Goldwyn, R. ~1984!. Predicting rejecting of Schneider-Rosen, K., Braunwald, K., Carlson, V., & Cic- her infant from mother’s representation of her own ex- chetti, D. ~1985!. Current perspectives in attachment perience: Implications for the abused–abusing inter- theory: Illustrations from the study of maltreated in- generational cycle. ChildAbuse and Neglect,8, 203–217. fants. Monographs of the Society for Research in Child Main, M., & Hesse, P. ~1990!. Parents’ unresolved trau- Development, 50, 194–210. matic experiences are related to infant disorganized Schore, A. N. ~2003!. Affect regulation and disorders of attachment status: Is frightened and0or frightening the self. New York: W.W. Norton. parent behavior the linking mechanism? In M. Green- Shirk, S. R. ~1999!. Integrated child psychotherapy: Treat- berg, D. Cicchetti, & E. M. Cummings ~Eds.!, Attach- ment ingredients in search of a recipe. In S. W. Russ ment in the preschool years ~pp. 161–182!. Chicago: & T. H. Ollendick ~Eds.!, Handbook of psycho- University of Chicago Press. therapies with children and families: Issues in clini- Fostering secure attachment in infants 649

cal child psychology ~pp. 369–384!. Dordrecht: Kluwer chopathology and child psychotherapy. In S. Russ & Academic. T. Ollendick ~Eds.!, Handbook of psychotherapies with Shirk, S. R., Talmi, A., & Olds, D. ~2000!. A develop- children and families ~pp. 15–44!. New York: Ple- mental psychopathology perspective on child and num Press. adolescent treatment policy. Development and Psycho- Toth, S. L., Maughan, A., Manly, J. T., Spagnola, M., & pathology, 12, 835–855. Cicchetti, D. ~2002!. The relative efficacy of two Smith, C. A., & Thornberry, T. ~1995!. The relationship interventions in altering maltreated preschool chil- between child maltreatment and adolescent involve- dren’s representational models: Implications for at- ment in delinquency. Criminology, 33, 451–481. tachment theory. Development and Psychopathology, Sroufe, L. A. ~1979!. The coherence of individual devel- 14, 777–808. opment: Early care, attachment, and subsequent de- United States Advisory Board on Child Abuse and Ne- velopmental issues. American Psychologist, 34, glect. ~1990!. Child abuse and neglect: Critical first 834–841. steps in response to a national emergency. Washing- Sroufe, L. A. ~1989!. Pathways to adaptation and malad- ton, DC: Department of Health and Human Services. aptation: Psychopathology as developmental devia- United States Congress. ~1988!. Healthy children: Invest- tion. In D. Cicchetti ~Ed.!, Rochester Symposium on ing in the future. Washington, DC: US Government Developmental Psychopathology: The emergence of a Printing Office. discipline ~Vol. 1, pp. 13–40!. Hillsdale, NJ: Erlbaum. van den Boom, D. C. ~1994!. The influence of tempera- Sroufe, L. A. ~1996!. Emotional development: The orga- ment and mothering on attachment and exploration: nization of emotional life in the early years. New An experimental manipulation of sensitive responsive- York: Cambridge University Press. ness among lower-class mothers with irritable infants. Sroufe, L. A., & Fleeson, J. ~1986!. Attachment and the Child Development, 65, 1457–1477. construction of relationships. In W. Hartup & Z. Ru- van den Boom, D. C. ~1995!. Do first-year intervention bin ~Eds.!, Relationships and development ~pp. 51– effects endure? Follow-up during toddlerhood for a 76!. Hillsdale, NJ: Erlbaum. sample of Dutch irritable infants. Child Development, Sroufe, L. A., & Fleeson, J. ~1988!. The coherence of 66, 1798–1816. family relationships. In R. A. Hinde & J. Stevenson- van IJzendoorn, M. H., Juffer, F., & Duyvesteyn, M. G. C. Hinde ~Eds.!, Relationships within families: Mutual ~1995!. Breaking the intergenerational cycle of inse- influences ~pp. 27–47!. Oxford: Oxford University cure attachment: A review of the effects of attachment- Press. based interventions on maternal sensitivity and infant Sroufe, L. A., & Rutter, M. ~1984!. The domain of devel- security. Journal of Child Psychology and Psychiatry, opmental psychopathology. Child Development, 55, 36, 225–248. 17–29. van IJzendoorn, M. H., Schuengel, C., & Bakermans- Sroufe, L. A., & Waters, E. ~1977!. Attachment as an Kranenburg, M. J. ~1999!. Disorganized attachment in organizational construct. Child Development, 48, early childhood: Meta-analyses of precursors concom- 1184–1199. itants, and sequelae. Development and Psychopathol- Thompson, R. ~1998!. Early sociopersonality develop- ogy, 11, 225–249. ment. In W. Damon & N. Eisenberg ~Eds.!, Handbook Vaux, A., Riedel, S., & Stewart, D. ~1987!. Modes of of child psychology ~5th ed., Vol. 3, pp. 25–104!. New social support: The Social Support Behaviors ~SS-B! York: Wiley. scale. American Journal of Community Psychology, Toth, S. L., & Cicchetti, D. ~1999!. Developmental psy- 15, 209–237.