Attachment in Institutionalized and Community Children in Romania
Attachment in Institutionalized and Community Children in Romania
Introduction
This study investigates attachment in institutionalized and community children aged 12–31 months in Bucharest, Romania.
Attachment was assessed through ratings of attachment behaviors and caregiver descriptions in a structured interview.
Children in institutions showed significant attachment disturbances.
Quality of caregiving in institutions was linked to the formation and organization of attachment.
Findings remained consistent even after controlling for cognitive level, perceived competence, and quantitative interaction ratings.
Ratings of attachment behavior in the Strange Situation and caregiver reports of Reactive Attachment Disorder (RAD) showed moderate convergence.
Children in institutions are at high risk for social and behavioral issues, including attachment disturbances.
Disturbances of attachment have been a key focus in research on the effects of institutionalization for over 50 years.
Early Studies of Institutionalized Children
Descriptive studies by Spitz (1945), Goldfarb (1945), Provence and Lipton (1962), and Wolkind (1974) documented aberrant behaviors in institutionalized children, later forming the basis for the clinical syndrome of Reactive Attachment Disorder (RAD).
Tizard's 1960s study in London residential nurseries is particularly important.
Examined 65 children in nurseries from birth or soon after.
By ages 2-4: 24 adopted, 15 returned to birth families, 26 remained in institutions.
At age 4, the 26 institutionalized children were assessed:
8 (30.8%) were emotionally withdrawn and unresponsive.
10 (38.4%) were indiscriminate, seeking attention from strangers as much as caregivers.
8 (30.8%) had a preferred attachment to a nursery caregiver.
The first two groups (emotionally withdrawn and indiscriminate) were foundational for defining the two types of RAD:
Emotionally withdrawn/inhibited.
Indiscriminately social/disinhibited.
These types are described in both DSM-IV-TR (American Psychiatric Association, 2000) and ICD-10 (World Health Organization, 1992).
Contemporary Studies
Smyke, Dumitrescu, and Zeanah (2002) studied RAD signs in Bucharest, finding more signs in institutionalized children compared to community children attending child care.
Children from different units within the same institution showed varying RAD signs.
Reducing the number of caregivers per child per week (from 17 to 4) significantly decreased signs of both emotionally withdrawn/inhibited and indiscriminately social/disinhibited RAD.
One aim of the present study is to replicate these findings in a larger sample of Romanian children in institutions.
Longitudinal Studies of Adopted Children
Longitudinal studies of children adopted from Romanian institutions show consistently high levels of indiscriminate or disinhibited attachment behavior.
Chisholm, Carter, Ames, and Morison (1995) and Chisholm (1998) found indiscriminate behavior to be a persistent social abnormality.
Indiscriminate ‘‘friendliness’’ was noted in children at a median of 11 and 39 months post-adoption.
Measures of indiscriminate friendliness did not correlate well with measures of attachment security.
Chisholm (1998) found that while security of attachment increased between 11 and 39 months post-adoption, indiscriminate friendliness did not decrease, especially in those adopted after 8 months of age.
A home-based separation-reunion procedure showed unusual, atypical insecure attachment patterns in most children adopted from Romania after 8 months.
Higher scores on one measure of ‘‘indiscriminate friendliness’’ were observed in insecurely attached children adopted from institutions.
O’Connor, Bredenkamp, Rutter, and the ERA Study Team (1999) and O’Connor, Rutter, and the ERA Study Team (2000) also identified indiscriminate behavior in children adopted from Romanian institutions into the UK, assessed at ages 4 and 6.
Longer duration of institutionalization was linearly related to signs of indiscriminate behavior at follow-up.
Convergence was found between parents' reports of disinhibited behavior and children’s attachment classifications in a home-based Strange Situation Procedure (O’Connor, Marvin, Rutter, Olrick, Britner, & ERA Study Team, 2003).
Children reported as highly indiscriminate were more likely to have Strange Situation Procedure classifications that were neither secure nor organized insecure.
In the Romanian sample, 73% of highly indiscriminate children were classified as insecure-disorganized/controlling or insecure-other, and 27.5% of children with these unusual classifications had highly indiscriminate behavior.
Another goal of the present study is to examine the convergence of SSP classifications and indiscriminate sociability in institutionalized children.
Emotionally Withdrawn/Inhibited RAD
Chisholm (1998) and O’Connor et al. (2000, 2003) did not describe the emotionally withdrawn/inhibited type of RAD, unlike Smyke et al. (2002).
Zeanah and Smyke (2002) suggested that adoption leads to attachment formation, reducing the absence of attachment characteristic of the emotionally withdrawn/inhibited type.
The current study aims to replicate and extend these findings by examining Strange Situation classifications in institutionalized children with high levels of emotionally withdrawn/inhibited RAD.
The Strange Situation Procedure in Institutional Settings
It is unclear how well the Strange Situation can detect the presence or absence of attachment in institutional settings, as it was developed to assess the quality of attachment (Ainsworth, Blehar, Waters, & Wall, 1978).
Vorria et al. (2003) found that 66% of Greek institutionalized infants had disorganized attachments, and 8% had unclassifiable attachments. Only 24% were securely attached.
In a comparison group of home-reared infants, 41% were securely attached, 25% were disorganized, and 22% were unclassifiable.
The authors noted that the comparison group may not have been typical, as they were recruited from child-care centers (unusual in Greece).
Birth weight, prematurity, health status, gender, ethnicity, temperament, cognitive abilities, nor caregiver sensitivity differentiated between secure and disorganized infants in the institutionalized group.
Another goal of the current study is to determine if patterns of attachment between institutionalized children and their caregivers can be assessed using the Strange Situation in Romanian institutions.
This is important because the Strange Situation assumes an attachment exists, which may not be the case in Romanian institutions where caregiver-to-child ratios are typically 1:12 (Smyke et al., 2002; Zeanah et al., 2003), compared to 1:4-6 in Greek institutions (Vorria et al., 2003).
Quality of Caregiving
The question of caregiving quality in institutions is significant.
Previous studies showed individual differences in children’s attachment behaviors, but only one tried to link it to caregiver behavior.
It's important to know if caregiver behavior is related to children’s attachment in a depriving environment.
Vorria et al. (2003) found no link between caregiver sensitivity and attachment security, but only assessed sensitivity from 20 minutes of free play.
The present study looks at caregiver behavior observed naturalistically in relation to children’s attachment in institutions.
Bucharest Early Intervention Project (BEIP)
Baseline data from the Bucharest Early Intervention Project (BEIP), a randomized controlled trial of foster care as an alternative to institutionalization, were used.
The study included all institutionalized children in Bucharest not scheduled for imminent international adoption or with severe handicaps (Zeanah et al., 2003).
Research Questions
Are there more signs of indiscriminately social/disinhibited and emotionally withdrawn/inhibited RAD in institutionalized children compared with community children?
Prediction: Both types of RAD will be more common in the institutionalized group, based on previous research (Smyke et al., 2002; Tizard & Rees, 1975; Chisholm, 1998; O’Connor et al., 2000, 2003).
Is it possible to identify patterns of attachment between young children in institutions and their ‘‘favorite’’ caregivers?
Prediction: Attachments can be demonstrated using the Strange Situation and caregiver reports.
Are disorganized attachments the predominant classification, and are there proportionately fewer organized patterns of attachment in institutionalized children compared with community children in Bucharest?
Prediction: Based on Vorria et al. (2003) and the less optimal caregiver-to-child ratio in Romanian institutions, the answers to both questions will be ‘‘yes.’’
Is there convergence between Strange Situation Procedure classifications of attachment and caregiver reports of signs of RAD?
Prediction: There will be convergence, although previous studies in post-institutionalized children have yielded mixed results (Chisholm, 1998; O’Connor et al., 2003).
Is the quality of caregiving related to classifications of attachment and/or to signs of RAD?
Prediction: Organized attachments will be associated with more sensitive/responsive caregivers in both groups. Lower caregiving quality will be associated with increased signs of both types of RAD.
Does cognitive impairment and/or perceived competence explain children’s attachment in the institutional setting?
Prediction: While both cognitive level and perceived competence will be related to attachment, neither will be sufficient to explain it.
Method
Participants
Two groups of children from the BEIP study (Zeanah et al., 2003) were included:
Institutionalized Group: 136 children who had spent an average of 90% of their lives in institutions in Bucharest.
Never Institutionalized Group: 72 Romanian children who had never been institutionalized, recruited from pediatric clinics.
Children were selected from the BEIP sample if they were at least 12 months of age and had a cognitive age of at least 11 months (derived from Bayley scores).
The final sample included 95 children in the institutional group and 50 children in the community group.
Proportionately more of the community group were chronologically younger at baseline.
Institutionalized Group:
52 boys and 43 girls.
Age range: 12 to 31 months (M = 23.85, SD = 4.85).
Never Institutionalized Group:
25 boys and 25 girls.
Age range: 12 to 31 months (M = 22.25, SD = 5.01).
Demographic data (Table 1) showed that the only difference between the groups was an overrepresentation of children of Roma ethnicity in the Institutionalized Group.
Measures
Attachment Quality: Assessed using the Strange Situation Procedure (Ainsworth et al., 1978).
Community children were assessed with their mothers.
Institutionalized children were assessed with their ‘‘favorite’’ caregivers, as determined by staff consensus.
If no favorite caregiver could be identified, the child was seen with a caregiver who worked regularly with the child.
Standard laboratory procedures were followed, with markers distinguishing group status eliminated (e.g., all caregivers dressed in street clothes).
Videotaped assessments were coded by expert raters (Elizabeth Carlson and Alan Sroufe) who were unaware of the children’s group status.
Dyads were assigned to one of four major classifications: secure (B), anxious avoidant (A), anxious resistant (C), and disorganized/disoriented (D) (Ainsworth et al., 1978; Main & Solomon, 1990).
Interrater reliability for classifications was adequate (k = .78). Differences were resolved by conferencing.
Attachment Formation Rating Scale:
A 5-point rating scale was developed to document the range of child behavior exhibited in the assessment that did not fit the traditional classification scheme, but appeared to reflect the degree of attachment formation (Ainsworth, 1967).
Ratings of ‘‘5’’ indicated attachment behavioral organization consistent with traditional A, B, C, and D classifications.
Ratings of ‘‘4’’ indicated evidence of attachment behavioral organization and the presence of extreme or pervasive behavioral anomalies (beyond the scope of traditional disorganization coding).
Ratings of ‘‘3,’’ ‘‘2,’’ and ‘‘1’’ were assigned for behavioral displays ranging from fragmented or incomplete sequences of attachment behavior differentially directed toward the caregiver, to isolated attachment signals and responses, to no evidence of attachment behavior.
Intraclass correlation coefficient for interrater reliability was .96 (n = 45).
Attachment Disorder
The Disturbances of Attachment Interview (Smyke & Zeanah, 1999) was used to assess signs of clinical disturbances of attachment.
The interview has shown to distinguish between institutionalized and never institutionalized children in Romania (Smyke et al., 2002; Zeanah, Smyke, & Dumitrescu, 2002) and to identify signs of RAD reliably in maltreated children (Zeanah et al., 2004).
It includes 12 items, each explored through a series of probes.
Items are rated as ‘‘0’’ = ‘‘none or never,’’ ‘‘1’’ = ‘‘somewhat or sometimes,’’ and ‘‘2’’ = ‘‘considerable or frequently.’’
Each interview was coded by two native Romanian coders, with discrepancies resolved by conferencing.
The first five items assess signs of emotionally withdrawn/inhibited RAD (scores range from 0 to 10).
The next three items assess signs of indiscriminately social/disinhibited RAD (scores range from 0 to 6).
Caregiving Environment
The Observational Record of the Caregiving Environment (ORCE; NICHD Child Care Research Network, 1996, 1997, 2003) was adapted to assess a specific child’s caregiving environment.
Subjects were videotaped in their environment for 1 1/2 hours, instead of using the live coding approach.
Qualitative items were added, such as marked dysregulation, stereotypical behavior, and communicative gesture.There were 55 items that assessed caregiver behavior (e.g., positive physical contact, asks questions of child) as well as child behavior (e.g., activity with objects, unoccupied/watching).
Coders watched the tape for 10 minutes and then began to code items at the first even time (e.g., 12:20:00).
Coders watched for a 30-second observation period and proceeded to mark behavioral items as present or not present during the subsequent 30-second coding period.
Qualitative items were then rated on a scale ranging from 1 (not at all characteristic) to 4 (highly characteristic).
Examples of qualitative items included caregiver detachment and sensitivity to child distress.
For the current sample, 40% of tapes were randomly selected for double coding to ensure the ongoing fidelity of the process (reliability = 95%).
Cognitive Abilities
The Bayley Scales of Infant Development II (BSID-II; Bayley, 1993) was used to assess the developmental status of study participants.
Administered at the baseline evaluation, except for one child from the Institutionalized Group who was ill.
The Mental Development Index (MDI) scaled score ranges from 50 to 150.
Children who obtained raw scores that placed them below 50 for their chronological age were assigned a numeric score of 49.
Extrapolated age scores were divided by chronological age and that amount was multiplied by 100 to produce a Developmental Quotient (DQ) for each participant.
Child Behavior Problems and Competence
The Infant Toddler Social Emotional Assessment (ITSEA; Carter & Briggs-Gowan, 2000) is a 195-item questionnaire that was administered to caregivers to assess a variety of problem behaviors and competencies in children 12–48 months of age.
Only the competence score, derived from six subscales (Attention, Compliance, Imitation/Play, Mastery Motivation, Empathy, Prosocial Peer Relations) was included for this report.
Internal consistency assessment of this measure suggested appropriate reliability for the Competence scale ().
Procedures
Children from the Institutionalized Group were recruited with the assistance of the National Authority for Child Protection and the Commissions for Child Protection.
Children from the Never Institutionalized Group were recruited from pediatric clinics administered by the Institute for Maternal and Child Health in Bucharest (see Zeanah et al., 2003).
Assessments, including Strange Situation Procedures and interviews, were conducted at the BEIP laboratory in Bucharest.
‘‘Home’’ observations were made in the institutional settings of the Institutionalized Group and in the homes of the Never Institutionalized Group, where 1 1/2 hours of ‘‘typical behavior’’ were videotaped and later coded.
Results
There were no associations between age, gender, or ethnicity of children and measures of attachment.
Signs of RAD
Institutionalized children (M = 1.92, SD = 2.04) had significantly higher levels of RAD emotionally withdrawn/inhibited ratings than Never Institutionalized children (M = .26, SD = 0.57), t(135) = 5.3, p < .001, d = .91.
Institutionalized children (M = 2.17, SD = 1.71) also received higher ratings of RAD indiscriminately social/disinhibited than children in the Never Institutionalized comparison group (M = 1.39, SD = 1.27), t(138) = 2.85, p < .01, d = .49.
These differences also were significant, RAD emotionally withdrawn/inhibited, \chi^2(2) = 42.74, p < .001, w = .51, and RAD indiscriminately social/disinhibited, \chi^2(2) = 10.86, p < .01, w = .47.
Within the Institutionalized Group, there was no apparent relationship between length of institutionalization and signs of either RAD emotionally withdrawn/inhibited (r = -.05, p > .05) or RAD indiscriminately social/disinhibited (r = -.074, p > .05).
Strange Situation Classifications
Only 22% of children in institutions had organized attachment strategies, compared to 78% of children living with their parents (\chi^2(1) = 43.283, p < .001, w = .54).
12.6% of the institutionalized group had so little attachment behavior that it could not even be classified as disorganized (designated as unclassifiable).
Continuous Ratings of Strange Situation Procedure Attachment Behavior
Every community child had an attachment rating of ‘‘5’’, while only 3/95 children in institutions had such a rating.
Except for one child rated as securely attached, all other organized attachment classifications in the Institutionalized Group were rated as less than a ‘‘5’’.
Disorganized attachments were all rated as ‘‘5’’ in the Never Institutionalized Group but ranged from ‘‘2’’ to ‘‘5’’ in the Institutionalized Group.
Convergence of Attachment Measures
Within the institutionalized group, continuous ratings of attachment were moderately correlated with caregiver ratings of RAD emotionally withdrawn/inhibited, r = -.44, p < .01, but unrelated to ratings of RAD indiscriminately social/disinhibited behavior, .
Having an organized attachment was not significantly related to caregiver ratings of signs of either type of RAD.
Quality of Caregiving and Classifications of Attachment
Within the Institutionalized Group, analysis of variance (ANOVA) was used to examine quality of caregiving with attachment status (organized vs. disorganized vs. unclassified) as the grouping variable.
Scores of children who fell in the Organized and Disorganized groups (M = 15.43, SD = 3.16; M = 14.41, SD = 3.59, respectively) were not different from one another but were greater than those of children in the Unclassified group (M = 11.54, SD = 2.55), F(1, 92) = 5.107, p < .01.
Post hoc Tukey’s testing indicated that the quality of caregiving received by the children in the Unclassified group was poorer than that received by the Organized group (p = .006) and by the Disorganized group (p = .024).
Within the Institutionalized Group, multiple linear regression was used to examine possible relationships between the continuous ratings of attachment and quality of caregiving, cognitive level, and competence measures.
After controlling for the effects of cognitive development, quantitative aspects of caregiver-child interaction, and competence score, the only measure that significantly predicted attachment rating in institutionalized children was quality of caregiving, .
Each unit increase in the quality of caregiving was associated with an increase in attachment rating of .119.
For purposes of determining the likelihood that a child in the institution had an organized attachment (as opposed to disorganized, with unclassified excluded), logistic regression was used with attachment rating, cognitive development, competence score, quantitative aspects of caregiver-child interaction, and quality of caregiving as possible factors.
After controlling for the effects of the aforementioned factors, attachment rating (Wald = 6.793, p = .009, df = 1) and quality of caregiving (Wald = 3.951, p = .047, df = 1) were the only significant factors.
An increase of 1 unit in attachment rating was associated with an increase in the odds of an institutionalized child having an organized attachment by a factor of .379.
An increase of 1 unit in quality of caregiving was associated with an increase in the odds of an institutionalized child having an organized attachment by a factor of 1.292.
When children from the Unclassified group were included in the disorganized group, and the same factors were used in the logistic regression, quality of caregiving was the only significant factor (Wald = 4.409, p = .036, df = 1).
An increase of one unit in quality of caregiving was associated with a 30% increase in the odds of an institutionalized child having an organized attachment.
Similar logistic regressions performed on the Never Institutionalized Group revealed that there appeared to be no differences in the odds of a child developing an organized attachment based on attachment rating, cognitive development, competence score, quantitative aspects of caregiver-child interaction, or quality of caregiving.
In the Institutionalized Group only, quality of caregiving was related to RAD emotionally withdrawn/inhibited scores, , but unrelated to RAD indiscriminately social/disinhibited scores, .
Discussion
This is the largest and most comprehensive study of attachment in institutionalized children to date.
It replicated and extended findings of previous studies.
Serious disturbances of attachment are the rule rather than the exception in Romanian institutions.
Institutionalized children demonstrated more signs of both types of RAD than community children.
This replicates the findings of Smyke et al. (2002) but with a larger and more representative sample.
Interestingly, there was no relationship between length of institutionalization and signs of RAD in the ages studied (11–31 months).
This likely results from the fact that the children are still in the adverse caregiving environment.
The levels of unclassifiable and disorganized attachment in this investigation are comparable with the only other study of institutionalized children using the Strange Situation Procedure (Vorria et al., 2003) and to a study of maltreated toddlers and their caregivers (Carlson, Cicchetti, Barnett, & Braunwald, 1989).
Even when discernible attachment patterns were observed, they appeared to be anomalous or incompletely developed.
The enormous discrepancy between the Never Institutionalized Group and Institutionalized group demonstrates this point.
These results strongly suggest that secure attachments in the Institutionalized Group are not comparable with secure attachment in the Never Institutionalized Group.
Similarly, even disorganized attachments in the Institutionalized Group and the Never Institutionalized Group were different.
A simple comparison of the distributions of attachment classifications underestimates the magnitude of the differences in attachment.
In the Stovall and Dozier sample, the children were in the early stages of constructing a relationship with a new caregiver, so that we would anticipate the behaviors becoming more robust over time.
In contrast, the institutionalized children we studied were in a chronic state of deprivation in which the amount and quality of contact that they received was unlikely to change significantly, barring some unusual development.
The results suggest that the majority of institutionalized children in this study were not able to form selective attachments to their caregivers.
Levels 4 and 5 are compatible with selected attachments, but Levels 3 and below are not.
Failure to examine this additional perspective may obscure important differences.
Longitudinal data are needed to determine the course and outcome of level of attachment formation.
The moderate association between caregiver reports of signs of emotionally withdrawn/inhibited RAD and ratings of the child’s behavior with the caregiver in the Strange Situation Procedure provides evidence of cross-validation for both the interview measure and the continuous rating of attachment behavior.
The fact that the convergence was only moderate also demonstrates that they provide somewhat different indices of attachment disturbance.
These results suggest that clinical disturbances, as reflected in signs of RAD, are related to how fully developed and expressed attachment behaviors are, but not necessarily to the organization of any particular pattern of attachment.
There was no relationship between caregiver ratings of indiscriminately social/disinhibited RAD and Strange Situation Procedure ratings of attachment behavior.
This fails to replicate the O’Connor et al. (2003) findings, which showed that aberrant patterns of attachment were more common in children who had high ratings of indiscriminate behavior.
Another important contribution of this study was the demonstration that the quality of the caregiving that the child received in the institutional setting was significantly related both to the continuous rating of attachment and to the child’s organization of attachment.
Impressively, these results held even when other variables, such as cognitive level, perceived competence, and quantitative interaction ratings, were controlled for.
In contrast, there was no relationship between caregiving quality and attachment in the community setting.
Variability in caregiving quality within the community sample may have been too limited to capture differences that related to attachment.
Limitations
Romanian institutions, which are characterized for the most part with particularly poor caregiver to child ratios, may not be representative of some institutions in other countries.
In this sample the average Bayley MDI score was 65 (see Zeanah, Smyke, & Koga, 2003).
These data are cross sectional, and a number of important questions await longitudinal follow-up.
The answer to the direction of effects question has important implications for considerations of resilience.
Differences in institutional caregiver reports of children’s attachment behaviors and parent reports of community children’s attachment behaviors may be because of differences in relationships between adults and children rather than to true differences in the children’s behavior.
Coders of interviews (DAIs) and ORCE coders could not be completely blind to children’s group status, although they were unaware of the design or hypotheses of the study.
Strange Situation coders were, however, blind to children’s group status.
Assessment of Strange Situation Procedure classifications in children between 20 and 30 months poses challenges.
Conclusion
Taken together with previous research on attachment and institutionalization, it is clear that attachment is a severely compromised developmental domain in young, institutionalized children.
The importance of quality caregiving for young children in extreme conditions of social deprivation is clear.
Caregiver’s sensitive responsiveness to children’s distress and active engagement with the children enhances the probability of formation of a more developed and more organized attachment.
Essential questions about the potential for recovery of attachment, how timing of intervention relates to recovery, and which factors enhance or impede recovery, remain to be addressed in future research.
Introduction
This study rigorously examines attachment patterns among children aged 12–31 months in Bucharest, Romania, contrasting those raised in institutional settings with those from community environments.
The assessment of attachment relies on detailed ratings of attachment behaviors and comprehensive caregiver descriptions gathered through structured interviews, ensuring a multifaceted evaluation.
The research highlights that children in institutions exhibit significant disturbances in attachment, evidencing compromised emotional and social development.
A key finding is the direct relationship between the quality of caregiving within institutions and the subsequent formation and organization of attachment patterns in children.
The study's findings are robust, maintaining consistency even after accounting for variables such as cognitive level, perceived competence, and quantitative interaction ratings, thereby isolating the impact of institutional care.
Moderate convergence was observed between ratings of attachment behaviors in the Strange Situation Procedure and caregiver reports of Reactive Attachment Disorder (RAD), validating the assessment methods.
It is underscored that children in institutional care are particularly vulnerable to social and behavioral challenges, notably attachment disturbances, which can have long-term consequences.
Research on the effects of institutionalization has consistently focused on disturbances of attachment for over half a century, recognizing its critical role in child development.
Early Studies of Institutionalized Children
Pioneering descriptive studies by Spitz (1945), Goldfarb (1945), Provence and Lipton (1962), and Wolkind (1974) meticulously documented aberrant behaviors in institutionalized children, which later served as the groundwork for defining the clinical syndrome of Reactive Attachment Disorder (RAD).
Tizard's 1960s study in London residential nurseries holds substantial importance due to its longitudinal examination of children's developmental outcomes.
The study involved 65 children in nurseries from birth or shortly thereafter, providing a detailed account of their early experiences.
By ages 2-4:
24 children were adopted, marking a transition to family-based care.
15 children were returned to their birth families, indicating a shift in their care arrangement.
26 children remained in institutions, continuing to experience institutionalized care.
At age 4, the 26 institutionalized children underwent thorough assessment:
8 children (30.8%) exhibited emotional withdrawal and unresponsiveness, indicative of inhibited attachment patterns.
10 children (38.4%) displayed indiscriminate behavior, readily seeking attention from strangers as much as from caregivers, showcasing a lack of selective attachment.
8 children (30.8%) demonstrated a preferred attachment to a specific nursery caregiver, suggesting the capacity for forming selective attachments even within the institutional setting.
The identification of emotionally withdrawn and indiscriminate behaviors in these children laid the foundation for delineating the two primary subtypes of RAD:
Emotionally withdrawn/inhibited type, characterized by inhibited emotional expression and social interaction.
Indiscriminately social/disinhibited type, marked by excessive sociability and a lack of selectivity in forming attachments.
These classifications are formally recognized and delineated in both DSM-IV-TR (American Psychiatric Association, 2000) and ICD-10 (World Health Organization, 1992), providing standardized diagnostic criteria.
Contemporary Studies
Smyke, Dumitrescu, and Zeanah (2002) conducted research on RAD signs in Bucharest, revealing a higher prevalence of these signs among institutionalized children compared to community children attending child care.
The study further demonstrated variability in RAD signs among children from different units within the same institution, highlighting the influence of micro-environmental factors on attachment outcomes.
Notably, reducing the number of caregivers per child per week—from 17 to 4—led to a significant decrease in signs of both emotionally withdrawn/inhibited and indiscriminately social/disinhibited RAD, underscoring the importance of caregiver stability.
A central objective of the present study is to validate and expand upon these findings through a larger-scale investigation involving a more extensive sample of Romanian children residing in institutions.
Longitudinal Studies of Adopted Children
Longitudinal studies tracking children adopted from Romanian institutions consistently report elevated levels of indiscriminate or disinhibited attachment behavior, highlighting the long-term impact of early institutionalization.
Chisholm, Carter, Ames, and Morison (1995) and Chisholm (1998) identified indiscriminate behavior as a persistent social abnormality, emphasizing its resistance to change over time.
Indiscriminate ‘‘friendliness’’ was observed in children at a median of 11 and 39 months post-adoption, indicating its presence both shortly after adoption and in the longer term.
Measures of indiscriminate friendliness exhibited limited correlation with measures of attachment security, suggesting that these constructs capture distinct aspects of social and emotional functioning.
Chisholm (1998) found that while security of attachment tended to increase between 11 and 39 months post-adoption, indiscriminate friendliness showed minimal reduction, particularly among children adopted after 8 months of age, highlighting a sensitive period for attachment development.
A home-based separation-reunion procedure revealed unusual, atypical insecure attachment patterns in the majority of children adopted from Romania after 8 months, underscoring the profound impact of early deprivation on attachment organization.
Higher scores on a measure of ‘‘indiscriminate friendliness’’ were observed in insecurely attached children adopted from institutions, reinforcing the link between indiscriminate behavior and compromised attachment security.
O’Connor, Bredenkamp, Rutter, and the ERA Study Team (1999) and O’Connor, Rutter, and the ERA Study Team (2000) also documented indiscriminate behavior in children adopted from Romanian institutions into the UK, with assessments conducted at ages 4 and 6, emphasizing the cross-cultural consistency of these findings.
A linear relationship was found between longer duration of institutionalization and increased signs of indiscriminate behavior at follow-up, indicating a cumulative effect of institutional care on social disinhibition.
Convergence was observed between parents' reports of disinhibited behavior and children’s attachment classifications in a home-based Strange Situation Procedure (O’Connor, Marvin, Rutter, Olrick, Britner, & ERA Study Team, 2003), providing further validation of the assessment methods.
Children reported as highly indiscriminate were more likely to exhibit Strange Situation Procedure classifications indicative of neither secure nor organized insecure attachment patterns, suggesting disruptions in attachment organization.
In the Romanian sample, 73% of highly indiscriminate children were classified as insecure-disorganized/controlling or insecure-other, and 27.5% of children with these unusual classifications displayed highly indiscriminate behavior, highlighting the co-occurrence of these disturbances.
Another key objective of the present study is to ascertain the degree of convergence between SSP classifications and indiscriminate sociability among institutionalized children, contributing to a more comprehensive understanding of attachment disturbances.
Emotionally Withdrawn/Inhibited RAD
Chisholm (1998) and O’Connor et al. (2000, 2003) did not specifically address the emotionally withdrawn/inhibited type of RAD in their investigations, contrasting with the findings of Smyke et al. (2002).
Zeanah and Smyke (2002) proposed that adoption facilitates attachment formation, thereby diminishing the absence of attachment characteristic of the emotionally withdrawn/inhibited type, suggesting a potential pathway to recovery.
The current study seeks to replicate and build upon these insights by examining Strange Situation classifications in institutionalized children exhibiting pronounced levels of emotionally withdrawn/inhibited RAD, shedding light on the interplay between attachment patterns and RAD subtypes.
The Strange Situation Procedure in Institutional Settings
The applicability of the Strange Situation Procedure (SSP) in detecting the presence or absence of attachment in institutional settings remains uncertain, as it was originally designed to evaluate the quality of attachment in typically developing children (Ainsworth, Blehar, Waters, & Wall, 1978).
Vorria et al. (2003) reported that 66% of Greek institutionalized infants exhibited disorganized attachments, and 8% had unclassifiable attachments, highlighting the prevalence of atypical attachment patterns in this population. Only 24% were classified as securely attached.
In a comparison group of home-reared infants, 41% were securely attached, 25% were disorganized, and 22% were unclassifiable, indicating substantial differences in attachment distributions between institutionalized and home-reared children.
The authors acknowledged that the comparison group may not have been representative, as they were recruited from child-care centers—a practice not typical in Greece—potentially influencing the findings.
Factors such as birth weight, prematurity, health status, gender, ethnicity, temperament, cognitive abilities, and caregiver sensitivity did not differentiate between secure and disorganized infants within the institutionalized group, suggesting that these variables may not fully account for attachment disorganization.
A significant aim of the current study is to assess the utility of the Strange Situation in evaluating attachment patterns between institutionalized children and their caregivers within Romanian institutions, addressing a critical methodological gap.
This is particularly important given that the Strange Situation assumes the existence of an attachment bond, which may be compromised in Romanian institutions characterized by caregiver-to-child ratios of approximately 1:12 (Smyke et al., 2002; Zeanah et al., 2003), compared to ratios of 1:4-6 in Greek institutions (Vorria et al., 2003).
Quality of Caregiving
The assessment of caregiving quality within institutional settings is of paramount importance for understanding attachment outcomes.
While prior studies have documented individual differences in children’s attachment behaviors, only limited research has explored the direct link between these behaviors and caregiver behavior, highlighting a gap in the literature.
Determining whether caregiver behavior is associated with children’s attachment patterns within a depriving environment is crucial for informing intervention efforts.
Vorria et al. (2003) found no significant association between caregiver sensitivity and attachment security, although sensitivity was only assessed during a brief 20-minute free play period, potentially limiting the scope of the assessment.
The present study aims to address this limitation by examining caregiver behavior observed naturalistically in relation to children’s attachment patterns within institutions, providing a more ecologically valid assessment.
Bucharest Early Intervention Project (BEIP)
The study utilizes baseline data from the Bucharest Early Intervention Project (BEIP), a randomized controlled trial designed to evaluate foster care as an alternative to institutionalization, offering a rigorous methodological framework.
The study encompassed all institutionalized children in Bucharest who were not slated for imminent international adoption or experiencing severe handicaps, ensuring a representative sample (Zeanah et al., 2003).
Research Questions
Are there more signs of indiscriminately social/disinhibited and emotionally withdrawn/inhibited RAD in institutionalized children compared with community children?
Prediction: Based on prior research (Smyke et al., 2002; Tizard & Rees, 1975; Chisholm, 1998; O’Connor et al., 2000, 2003), both types of RAD are expected to be more prevalent in the institutionalized group, reflecting the adverse effects of institutional care.
Is it possible to identify patterns of attachment between young children in institutions and their ‘‘favorite’’ caregivers?
Prediction: The study anticipates that attachments can be identified and assessed using the Strange Situation Procedure and caregiver reports, indicating the potential for attachment formation even in institutional settings.
Are disorganized attachments the predominant classification, and are there proportionately fewer organized patterns of attachment in institutionalized children compared with community children in Bucharest?
Prediction: Drawing on the findings of Vorria et al. (2003) and considering the less favorable caregiver-to-child ratio in Romanian institutions, the study predicts that disorganized attachments will be predominant and that organized patterns of attachment will be less frequent among institutionalized children.
Is there convergence between Strange Situation Procedure classifications of attachment and caregiver reports of signs of RAD?
Prediction: Convergence between these measures is expected, although previous studies involving post-institutionalized children have yielded mixed results (Chisholm, 1998; O’Connor et al., 2003), warranting careful examination.
Is the quality of caregiving related to classifications of attachment and/or to signs of RAD?
Prediction: The study anticipates that organized attachments will be associated with more sensitive/responsive caregivers in both groups, while lower caregiving quality will be linked to increased signs of both types of RAD, underscoring the role of caregiver behavior in attachment outcomes.
Does cognitive impairment and/or perceived competence explain children’s attachment in the institutional setting?
Prediction: While both cognitive level and perceived competence are expected to be related to attachment, the study posits that neither will fully account for the observed attachment patterns, highlighting the importance of other factors.
Method
Participants
The study included two distinct groups of children from the BEIP study (Zeanah et al., 2003), allowing for a rigorous comparison:
Institutionalized Group: Comprising 136 children who had spent an average of 90% of their lives in institutions in Bucharest, representing the target population of interest.
Never Institutionalized Group: Consisting of 72 Romanian children who had never experienced institutionalization, recruited from pediatric clinics to serve as a community-based comparison group.
Children were selected from the BEIP sample based on the criteria of being at least 12 months of age and having a cognitive age of at least 11 months (as determined by Bayley scores), ensuring the inclusion of children within a specific developmental range.
The final sample consisted of 95 children in the institutional group and 50 children in the community group, reflecting the study's sample size and group composition.
It is noted that proportionately more children in the community group were chronologically younger at baseline, potentially influencing the interpretation of findings.
Institutionalized Group:
Included 52 boys and 43 girls, providing a balanced gender representation within the institutionalized sample.
The age range spanned from 12 to 31 months (M = 23.85, SD = 4.85), indicating the age distribution of the institutionalized children.
Never Institutionalized Group:
Consisted of 25 boys and 25 girls, ensuring equal gender representation in the community-based comparison group.
The age range was also 12 to 31 months (M = 22.25, SD = 5.01), aligning with the age range of the institutionalized group for comparability.
Demographic data (Table 1) revealed that the primary difference between the groups was an overrepresentation of children of Roma ethnicity in the Institutionalized Group, potentially introducing a confounding factor to consider in the analysis.
Measures
Attachment Quality:
Assessed using the Strange Situation Procedure (Ainsworth et al., 1978), a standardized protocol for evaluating attachment security in young children.
Community children were assessed with their mothers, reflecting the typical attachment relationship in home-reared children.
Institutionalized children were assessed with their ‘‘favorite’’ caregivers, as identified by staff consensus, to approximate the primary attachment figure within the institutional context.
In cases where a favorite caregiver could not be identified, the child was assessed with a caregiver who regularly interacted with the child, ensuring some level of familiarity in the assessment setting.
Standard laboratory procedures were rigorously followed, with measures taken to eliminate markers that could distinguish group status (e.g., all caregivers dressed in street clothes), minimizing potential bias.
Videotaped assessments were subjected to expert coding by trained raters (Elizabeth Carlson and Alan Sroufe) who were unaware of the children’s group status, ensuring objectivity in the coding process.
Dyads were assigned to one of four major classifications: secure (B), anxious avoidant (A), anxious resistant (C), and disorganized/disoriented (D) (Ainsworth et al., 1978; Main & Solomon, 1990), providing a comprehensive categorization of attachment patterns.
Interrater reliability for classifications was deemed adequate (k = .78), and discrepancies were resolved through conferencing to ensure consistency in coding.
Attachment Formation Rating Scale:
A custom-designed 5-point rating scale was employed to comprehensively document the range of child behaviors displayed during the assessment, including those that did not align with traditional classification schemes but provided insights into the degree of attachment formation (Ainsworth, 1967).
Ratings were assigned based on the following criteria:
Ratings of ‘‘5’’ indicated attachment behavioral organization consistent with traditional A, B, C, and D classifications, representing well-formed attachment patterns.
Ratings of ‘‘4’’ indicated evidence of attachment behavioral organization alongside extreme or pervasive behavioral anomalies that extended beyond the scope of traditional disorganization coding, capturing more complex attachment disturbances.
Ratings of ‘‘3,’’ ‘‘2,’’ and ‘‘1’’ were assigned to behavioral displays ranging from fragmented or incomplete sequences of attachment behavior directed toward the caregiver to isolated attachment signals and responses, and ultimately, to the absence of any discernible attachment behavior, reflecting varying degrees of attachment formation.
The intraclass correlation coefficient for interrater reliability was high at .96 (n = 45), ensuring the reliability of the rating scale.
Attachment Disorder
The Disturbances of Attachment Interview (Smyke & Zeanah, 1999) was administered to comprehensively assess signs indicative of clinical disturbances of attachment, a well-validated instrument in the field.
The interview has demonstrated its ability to differentiate between institutionalized and never institutionalized children in Romania (Smyke et al., 2002; Zeanah, Smyke, & Dumitrescu, 2002) and to reliably identify signs of RAD in maltreated children (Zeanah et al., 2004), establishing its validity and reliability.
The Disturbances of Attachment Interview (DAI) consists of 12 items, each thoroughly explored through a series of standardized probes, ensuring a systematic evaluation.
Items are rated on a 3-point scale: ‘‘0’’ = ‘‘none or never,’’ ‘‘1’’ = ‘‘somewhat or sometimes,’’ and ‘‘2’’ = ‘‘considerable or frequently,’’ enabling a nuanced assessment of the presence and frequency of attachment-related behaviors.
Each interview was independently coded by two native Romanian coders, and discrepancies were meticulously resolved through conferencing to ensure the accuracy and consistency of the coding process.
The first five items specifically target signs of emotionally withdrawn/inhibited RAD, yielding scores ranging from 0 to 10.
The subsequent three items focus on evaluating signs of indiscriminately social/disinhibited RAD, generating scores ranging from 0 to 6.
Caregiving Environment
The Observational Record of the Caregiving Environment (ORCE; NICHD Child Care Research Network, 1996, 1997, 2003) was adapted to comprehensively assess the specific caregiving environment of each child, providing valuable contextual information.
The ORCE differs from the live coding approach which instead involved videotaping participants in their natural environment for 1 1/2 hours, offering a more comprehensive and ecologically valid assessment.
Qualitative items were incorporated into the ORCE, including items related to marked dysregulation, stereotypical behavior, and communicative gestures, to capture additional nuances in the caregiving environment.The ORCE encompasses a total of 55 items that evaluate various facets of caregiver behavior (e.g., positive physical contact, asking questions of the child) and child behavior (e.g., activity with objects, unoccupied/watching). These items provide a multifaceted view of the dynamics within the caregiving setting.
The coding protocol involved coders watching the videotaped session for 10 minutes initially and then commencing the coding process at the first even time point (e.g., 12:20:00) to ensure standardized timing.
The coders watched for a 30-second observation period and subsequently marked behavioral items as either present or not present during the ensuing 30-second coding period, enabling a detailed and systematic assessment of observed behaviors.
Qualitative items were assessed using a rating scale ranging from 1 (not at all characteristic) to 4 (highly characteristic), allowing for the capture of subjective impressions regarding the caregiving environment.
Examples of qualitative items included measures of caregiver detachment and sensitivity to child distress, capturing essential dimensions of caregiving quality.
Rigorous measures were implemented to ensure ongoing fidelity in the coding process, with 40% of tapes randomly selected for double coding, resulting in a high level of reliability (95%).
Cognitive Abilities
The Bayley Scales of Infant Development II (BSID-II; Bayley, 1993) was utilized to assess the developmental status of the study participants, providing a standardized measure of cognitive functioning.
The BSID-II was administered during the baseline evaluation, except for one child from the Institutionalized Group who was ill at the time.
The Mental Development Index (MDI) scaled score ranges from 50 to 150, offering a reliable metric for gauging cognitive development.
For children who obtained raw scores that placed them below 50 for their chronological age, a numeric score of 49 was assigned, following established protocols.
Extrapolated age scores were divided by chronological age, and that result was multiplied by 100 to compute a Developmental Quotient (DQ) for each participant, enabling a standardized comparison of cognitive development across individuals.
Child Behavior Problems and Competence
The Infant Toddler Social Emotional Assessment (ITSEA; Carter & Briggs-Gowan, 2000) is a comprehensive 195-item questionnaire administered to caregivers to evaluate a range of problem behaviors and competencies in children aged 12–48 months, providing a holistic view of socio-emotional functioning.
For the purposes of this study, only the competence score, derived from six subscales (Attention, Compliance, Imitation/Play, Mastery Motivation, Empathy, Prosocial Peer Relations), was included in the analysis.
Internal consistency assessment of this measure indicated appropriate reliability for the Competence scale ().
Procedures
Children from the Institutionalized Group were recruited with the support of the National Authority for Child Protection and the Commissions for Child Protection to facilitate access to the target population.
Children from the Never Institutionalized Group were recruited from pediatric clinics administered by the Institute for Maternal and Child Health in Bucharest (see Zeanah et al., 2003) to ensure the inclusion of a community-based comparison group.
Assessments, involving Strange Situation Procedures and interviews, were conducted at the BEIP laboratory in Bucharest, providing a standardized and controlled setting for data collection.
‘‘Home’’ observations were conducted both in the institutional settings of the Institutionalized Group and in the homes of the Never Institutionalized Group, where 1 1/2 hours of ‘‘typical behavior’’ were videotaped and subsequently coded to capture the naturalistic dynamics of each environment.
Results
Statistical analyses revealed no significant associations between children's age, gender, or ethnicity and measures of attachment, suggesting that these demographic factors did not substantially influence attachment outcomes in this sample.
Signs of RAD
Institutionalized children (M = 1.92, SD = 2.04) exhibited significantly higher levels of RAD emotionally withdrawn/inhibited ratings than Never Institutionalized children (M = .26, SD = 0.57), t(135) = 5.3, p < .001, d = .91, indicative of a substantial difference in the presence of this subtype of RAD.
Institutionalized children (M = 2.17, SD = 1.71) also received higher ratings of RAD indiscriminately social/disinhibited than children in the Never Institutionalized comparison group (M = 1.39, SD = 1.27), t(138) = 2.85, p < .01, d = .49, demonstrating a noteworthy distinction in the manifestation of this RAD subtype as well.
These differences were further supported by chi-square analyses, which confirmed significant group effects for both RAD emotionally withdrawn/inhibited, \chi^2(2) = 42.74, p < .001, w = .51, and RAD indiscriminately social/disinhibited, \chi^2(2) = 10.86, p < .01, w = .47.
Within the Institutionalized Group, no apparent relationship was observed between the length of institutionalization and signs of either RAD emotionally withdrawn/inhibited (r = -.05, p > .05) or RAD indiscriminately social/disinhibited (r = -.074, p > .05), suggesting that duration alone did not fully account for the observed differences.
Strange Situation Classifications
A striking finding emerged: Only 22% of children in institutions exhibited organized attachment strategies, compared to a substantial 78% of children living with their parents (\chi^2(1) = 43.283, p < .001, w = .54), underscoring the profound impact of institutionalization on attachment organization.
Furthermore, 12.6% of the institutionalized group displayed such limited attachment behavior that it could not even be classified as disorganized (designated as unclassifiable), reflecting the severe disruption in attachment formation in this subset of children.
Continuous Ratings of Strange Situation Procedure Attachment Behavior
A notable distinction was observed: Every child in the community group received an attachment rating of ‘‘5’’, whereas only 3 out of 95 children in institutions attained this rating, highlighting the marked disparity in attachment quality.
With the exception of one child classified as securely attached, all other organized attachment classifications in the Institutionalized Group were rated as less than a ‘‘5’’, suggesting a weakened or compromised expression of attachment behaviors.
In the Never Institutionalized Group, disorganized attachments uniformly received a rating of ‘‘5’’, while in the Institutionalized Group, disorganized attachments varied in rating from ‘‘2’’ to ‘‘5’’, indicating a wider range of severity in the expression of disorganization.
Convergence of Attachment Measures
Within the institutionalized group, continuous ratings of attachment exhibited a moderate correlation with caregiver ratings of RAD emotionally withdrawn/inhibited, r = -.44, p < .01, suggesting a degree of concordance between these measures.
However, continuous ratings of attachment were unrelated to ratings of RAD indiscriminately social/disinhibited behavior, , indicating a lack of association between these constructs.
The presence of an organized attachment was not significantly related to caregiver ratings of signs of either type of RAD, suggesting that the organization of attachment alone did not fully account for the manifestation of RAD symptoms.
Quality of Caregiving and Classifications of Attachment
Within the Institutionalized Group, an analysis of variance (ANOVA) was conducted to evaluate the relationship between the quality of caregiving and attachment status (organized vs. disorganized vs. unclassified), providing a group-level comparison.
The results revealed that the scores of children who fell into the Organized and Disorganized groups (M = 15.43, SD = 3.16; M = 14.41, SD = 3.59, respectively) were not significantly different from each other but were significantly higher than those of children in the Unclassified group (M = 11.54, SD = 2.55), F(1, 92) = 5.107, p < .01, indicating a distinction in caregiving quality based on attachment classification.
Post hoc Tukey’s testing indicated that the quality of caregiving received by the children in the Unclassified group was significantly poorer than that received by the Organized group (p = .006) and the Disorganized group (p = .024), highlighting the association between lower caregiving quality and attachment disorganization.
Further, within the Institutionalized Group, multiple linear regression was employed to explore the relationships between continuous ratings of attachment and several factors, namely quality of caregiving, cognitive level, and competence measures.
After accounting for the effects of cognitive development, quantitative aspects of caregiver-child interaction, and competence score, the only measure that significantly predicted attachment rating in institutionalized children was quality of caregiving, , emphasizing the independent influence of caregiving quality on attachment outcomes.
Each unit increase in the quality of caregiving was associated with an increase in attachment rating of .119, demonstrating a quantifiable relationship between caregiving quality and attachment.
To determine the likelihood of a child in the institution having an organized attachment (as opposed to disorganized, excluding the unclassified group), a logistic regression was conducted with attachment rating, cognitive development, competence score, quantitative aspects of caregiver-child interaction, and quality of caregiving as possible factors.
After controlling for the effects of the aforementioned factors, attachment rating (Wald = 6.793, p = .009, df = 1) and quality of caregiving (Wald = 3.951, p = .047, df = 1) emerged as the only significant factors, underscoring their independent contributions to attachment organization.
An increase of 1 unit in attachment rating was associated with an increase in the odds of an institutionalized child having an organized attachment by a factor of .379, indicating the strength of this relationship.
Similarly, an increase of 1 unit in quality of caregiving was associated with an increase in the odds of an institutionalized child having an organized attachment by a factor of 1.292, further emphasizing the importance of caregiving quality.
When children from the Unclassified group were included in the disorganized group, and the same factors were used in the logistic regression, quality of caregiving remained the sole significant factor (Wald = 4.409, p = .036, df = 1).
In this analysis, an increase of one unit in quality of caregiving was associated with a 30% increase in the odds of an institutionalized child having an organized attachment, further supporting the crucial role of caregiving quality.
Conversely, similar logistic regressions conducted on the Never Institutionalized Group revealed that there appeared to be no significant differences in the odds of a child developing an organized attachment based on attachment rating, cognitive development, competence score, quantitative aspects of caregiver-child interaction, or quality of caregiving.
Finally, within the Institutionalized Group only, quality of caregiving was found to be related to RAD emotionally withdrawn/inhibited scores, , whereas it was unrelated to RAD indiscriminately social/disinhibited scores, , suggesting differential relationships between caregiving quality and specific RAD subtypes.
Discussion
This study represents the largest and most comprehensive investigation of attachment in institutionalized children conducted to date, providing valuable insights into this vulnerable population.
The study successfully replicated and expanded upon the findings of prior research, reinforcing the evidence base in this field.
A key takeaway is that serious disturbances of attachment are more the norm than the exception among children residing in Romanian institutions.
Institutionalized children exhibited a greater array of signs indicative of both subtypes of RAD compared to their community-dwelling counterparts.
This finding corroborates the work of Smyke et al. (2002) while leveraging a larger and more representative sample, thus enhancing the generalizability of the results.
Intriguingly, there was no discernible relationship between the duration of institutionalization and the manifestation of RAD signs within the age range examined (11–31 months).
This lack of association likely stems from the fact that the children remained within the same adverse caregiving environment throughout the study period.
The levels of both unclassifiable and disorganized attachment observed in this investigation are consistent with those reported in the only other study of institutionalized children employing the Strange Situation Procedure (Vorria et al., 2003) and in a study focusing on maltreated toddlers and their caregivers (Carlson, Cicchetti, Barnett, & Braunwald, 1989).
Notably, even when discernible attachment patterns were evident, they frequently appeared to be anomalous or incompletely developed.
The profound discrepancy observed between the Never Institutionalized Group and the Institutionalized group serves to underscore this critical point.
These results strongly imply that what might be designated as secure attachments within the Institutionalized Group cannot be considered equivalent to secure attachments observed in the Never Institutionalized Group.
Likewise, even disorganized attachments exhibited differences between the Institutionalized Group and the Never Institutionalized Group.
Consequently, a simple comparison of the distributions of attachment classifications is likely to underestimate the true magnitude of the differences in attachment.
In the Stovall and Dozier sample, the children were in the early stages of constructing a relationship with a new caregiver, so that we would anticipate the behaviors becoming more robust over time.
In contrast, the institutionalized children we studied were in a chronic state of deprivation in which the amount and quality of contact that they received was unlikely to change significantly, barring some unusual development.
The findings indicate that most institutionalized children in this study were not able to form selective attachments to their caregivers.
Levels 4 and 5 are compatible with selected attachments, but Levels 3 and below are not.
Failure to examine this additional perspective may obscure important differences.
Longitudinal data are needed to determine the course and outcome of level of attachment formation.
The moderate association between caregiver reports of signs of emotionally withdrawn/inhibited RAD and ratings of the child’s behavior with the caregiver in the Strange Situation Procedure provides evidence of cross-validation for both the interview measure and the continuous rating of attachment behavior.
The fact that the convergence was only moderate also demonstrates that they provide somewhat different indices of attachment disturbance.
These results suggest that clinical disturbances, as reflected in signs of RAD, are related to how fully developed and expressed attachment behaviors are, but not necessarily to the organization of any particular pattern of attachment.
There was no relationship between caregiver ratings of indiscriminately social/disinhibited RAD and Strange Situation Procedure ratings of attachment behavior.
This fails to replicate the O’Connor et al. (2003) findings, which showed that aberrant patterns of attachment were more common in children who had high ratings of indiscriminate behavior.
Another important contribution of this study was the demonstration that the quality of the caregiving that the child received in the institutional setting was significantly related both to the continuous rating of attachment and to the child’s organization of attachment.
Impressively, these results held even when other variables, such as cognitive level, perceived competence, and quantitative interaction ratings, were controlled for.
In contrast, there was no relationship between caregiving quality and attachment in the community setting.
Variability in caregiving quality within the community sample may have been too limited to capture differences that related to attachment.
Limitations
It is important to acknowledge that Romanian institutions, which are characterized for the most part by particularly poor caregiver-to-child ratios, may not be fully representative of all institutions in other countries, thus potentially limiting the generalizability of the findings.
In this sample, the average Bayley MDI score was 65 (see Zeanah, Smyke, & Koga, 2003), indicating that the children in this study exhibited lower cognitive development than typically developing children, which may influence the interpretation of the results.
The data presented are cross-sectional in nature, precluding the determination of causal relationships and longitudinal trends; thus, numerous questions warrant further investigation through longitudinal follow-up studies.
Addressing the question of the direction of effects holds critical implications for considerations