The impact of institutionalization on child development

Development and Psychopathology, 15 (2003), 853–884 Copyright  2003 Cambridge University Press Printed in the United States of America DOI: 10.1017.S0954579403000415 The impact of institutionalization on child development KIM MACLEAN St. Francis Xavier University Abstract During the past 10 years researchers studying children adopted from Romanian orphanages have had the opportunity to revisit developmental questions regarding the impact of early deprivation on child development. In the present paper the effects of deprivation are examined by reviewing both the early and more recent literature on studies of children who spent the first few years of life in institutions. Special attention is given to the Canadian study of Romanian adoptees in which the author has been involved. Findings across time and studies are consistent in showing the negative impact of institutionalization on all aspects of children’s development (intellectual, physical, behavioral, and social–emotional). Results of studies show, however, that institutionalization, although a risk factor for less optimal development, does not doom a child to psychopathology. However, the impact of institutionalization is greater when coupled with risk factors in the postinstitutional environment. Methodological and conceptual difficulties in research with institutionalized samples of children are discussed and future directions for research are considered. The attempt to understand human develop- differ from normal development. Longitudinal ment is greatly constrained by ethical consid- studies of previously institutionalized samples erations that make it impossible to examine can address questions concerning whether depthe impact of deprivation on child develop- rivation results in some skills never fully dement. “Experiments in nature” afforded by veloping or whether skills are simply delayed. samples of institutionalized children permit an There is also the opportunity to examine examination of the developmental process whether the behavior of orphanage children is when conditions are so severe as to impede quantitatively or qualitatively different from normal development. Findings from studies of the behavior of noninstitutionalized samples institutionalized children provide invaluable of children. information that can inform theory, research, Institutionalized samples also provide a and social policy with respect to both normal rare chance to examine how the roots of parand atypical development. Orphanage samples ticular pathologies may lie in behavior that permit us to evaluate developmental progress emerges as a result of the environmental conafter deprivation and examine how this might text in which children reside. The study of these behaviors has the potential to inform our thinking with respect to the roots of pathology Thanks to Elinor W. Ames and Megan Gunnar for their in other clinical samples of children. Revery helpful and thoughtful comments on earlier drafts of searchers can also examine the ways in which this manuscript. The author has previously published un- institutionalized samples are similar to and der the name Kim Chisholm. different from other samples of maltreated Address correspondence and reprint requests to: Kim MacLean, Psychology Department, St. Francis Xavier children. Children from orphanages usually experience maltreatment and neglect on every University, PO Box 5000, Antigonish, Nova Scotia, Canada; E-mail: [email protected]. level (physical, behavioral, social, and emo853 854 tional), but unlike many maltreated groups, they have experienced the intervention of adoption into supportive home environments that provide for developmental needs. Examination of these groups and their progress may suggest ways in which we might intervene with other maltreated samples. There are, however, inherent limitations to what can be learned from any one study of institutionalized children. Such experiments in nature are often considered to be real-life counterparts of experimental animal studies of early deprivation. However, it might be preferable, when using the phrase, to place more emphasis on the word nature and less on experiment. Whereas variables in experiments are isolated and controlled, this is not the case in real life. For example, for experimental purposes it is possible to restrict animals’ perceptual experience while keeping the rest of their environment equivalent to that of animals with unrestricted perceptual experience. For institutionalized children, on the other hand, the variable of “institutionalization” refers to a complex mix of social, perceptual, physical, intellectual, and emotional deprivation. As a result, when institution-reared children are found to differ from family-reared children, it is usually not possible to specify the type of deprivation that produced the differences. In institution studies the ages of entry into and exit from the institution are not under experimental control. Family needs, children’s conditions, and society’s threshold for removing children from their homes are all factors in determining entry. The age of leaving the institution varies according to national social and economic policy and the child’s health, physical attractiveness, and gender. When children enter institutions at or shortly after birth, their age at adoption is confounded with duration of deprivation, so that it is never possible to tell whether age of adoption or length of institutionalization is the effective variable. In the minority of cases in which children enter institutions considerably later than birth, the conditions under which they were raised before entering the institution are also relevant to the outcome. In any topic studied over a period of 60 years the use of different measures becomes a problem. Attachment problems in institution- K. MacLean alized children were first noted before the study of attachment began; indeed, it was studies of orphanage children that helped to lead Bowlby to formulate the concept of attachment (Bowlby, 1951). There is a question whether current measures of attachment are appropriate for assessing attachment in orphanage samples because they were not designed to evaluate the presence or absence of an attachment relationship. Rather, these measures assess the quality of attachment and make the assumption that an attachment relationship exists. The measurement of attachment is still not agreed upon; even recent studies of orphanage children’s attachment to their adoptive parents have used different systems to measure attachment at the preschool age. Another problem is the need to develop new measures for concepts described but not measured earlier. The indiscriminate friendliness toward adults often shown by previously institutionalized children was first described 58 years ago (Goldfarb, 1945b), but it is only in the last 8 years that researchers have attempted to develop measures; again, there is no one agreed-upon measure. Although the lack of agreement on measurement systems complicates the comparison of studies, the use of different systems not only broadens the number of different behaviors examined but may also clarify the meaning of such behavior. Institutionalization studies also have the problem of having to use measures standardized on home-reared children to assess institution-reared children. For example, many items on tests of capability at young ages are simply not applicable to orphanage populations. One does not often find a home-reared child who has never had an opportunity to grasp an object small enough to require finger involvement, but this is common in children who have been kept in cribs for the first year or more. Problems may arise with subscales on standardized tests of behavior problems: Can the same items that have been shown to fall into factors in nonclinical, or even clinical, samples be assumed to be organized into the same factors in a sample reared under such different conditions? The question of appropriate comparison groups for children in or from institutions is a thorny one. There are many possible com- Institutionalization and child development 855 parison groups, but each answers a different The point to be made is that there is no question and fails to answer other questions. perfect comparison group in nature that can Previously institutionalized children are both possibly answer all the important questions adopted and formerly institutionalized. To sep- that can be asked. Each of the different comarate the effects of institutionalization from parison groups teaches us something, but none those of adoption, for example, one might is a complete control group. Both the relook at adoptees who did not come from insti- searcher and the reader of research must pay tutions. Rutter and colleagues (Rutter & the careful attention to what a particular compariEnglish and Romanian Adoptees [ERA] Study son group can or cannot tell us. There is also Team, 1998) employed a comparison group an advantage to this situation, however. The of children adopted in the United Kingdom; fact that different studies have used different but all of these children were adopted before measures or types of comparison groups some6 months of age, which makes them an ade- times makes comparing them more difficult, quate comparison only for children adopted but it also increases the number of questions from Romania before age 6 months. Because that can be asked. In addition, when similar there are few children who go directly from conclusions are drawn from studies using difthe birth home to an adoptive home at later ferent measures and comparison groups, those ages, the researcher who wanted to compare conclusions may be accepted with considergroups of children who have been adopted able confidence. later would have to compare previously instiThis article reviews the impact of institututionalized children to domestic adoptees who tionalization on the following areas of child may have been abused or neglected in their development: physical development; developbirth homes and/or then moved through (some- mental milestones, intellectual development, times several) foster homes before adoption. and academic achievement; behavior probAlternatively, one could use a comparison lems; indiscriminate friendliness; and attachgroup of children who stay in institutions. ment. Under each area the characteristics of This is perfectly legitimate but it answers a institutionalized and previously institutionaldifferent question: how much better off are ized children are reviewed, especially with adoptees from institutions than those left be- regard to their differences from comparison hind? It does not tell us anything about how groups. Evidence on the persistence of or previously institutionalized children’s perfor- changes in the behavior, its relation to length mance compares to that of the home-reared of institutionalization, and its correlates are children among whom they now live. also presented. The results of recent studies of Canadian researchers (Ames, 1997) have adoptees from Romanian orphanages, espeused two different age-matched groups to com- cially the Canadian study in which this author pare to their orphanage group. One is an early- has been involved (Ames & Chisholm, 2001; adopted group of children who would have Chisholm, 1998, 2000; Chisholm, Carter, Ames, gone to Romanian orphanages had they not & Morison, 1995; Fisher, Ames, Chisholm, & been adopted before 4 months of age. This Savoie, 1997; Gunnar, Morison, Chisholm, & group has thus had the same poor prenatal and Shuder, 2001; Morison, Ames, & Chisholm, perinatal experience as institutionalized chil- 1995) are emphasized, but pioneer studies of dren but differs from the institutionalized group institutionalization are also referenced and in terms of age at adoption and (at any com- their conclusions included wherever relevant mon age) length of time spent in the adoptive to the more modern work. home. The second comparison group in the Canadian study consists of Canadian-born Developmental Milestones, Intellectual children reared in homes similar to those of Development, and Academic Achievement the orphanage children after adoption; however, in addition to being different with regard Research interest in the developmental conseto the crucial variable of orphanage experi- quences of extreme deprivation in infancy beence, the two groups also differ in the pres- gan intensely in the 1940s and 1950s with the ence or absence of the adoption experience. work of Rene Spitz (1945a, 1945b), William 856 Goldfarb (1943a, 1944, 1945a, 1947, 1955), and John Bowlby (1953). Most research focused on intellectual development. Rene Spitz (1945a, 1945b), who studied children in institutions, reported a drastic drop in infants’ developmental quotients (DQs) over the early months of institutional care. He reported that by the end of the second year infants’ DQs had dropped to a low of 45, as compared to an average DQ of 100. Spitz described the deterioration of children as progressive in spite of the fact that physical conditions in orphanage had improved over the course of his study. Given that improvements to orphanage infrastructure had not resulted in an improvement in DQ scores, Spitz concluded that children were irreparably damaged by institutionalization in the first years of life. William Goldfarb (1945a, 1955) studied 15 children who had been reared in an institution for the first 3 years of their lives and were subsequently placed in foster care. He compared those children to a group of children who had been in foster care since early infancy. Goldfarb found that even in adolescence the institution group was delayed intellectually relative to the foster care group, and he claimed that early institutional rearing resulted in developmental deficits that were not overcome once children were placed in more stimulating and loving environments. Researchers conducting this early work claimed quite strongly that children were intellectually compromised as a result of early institutionalization. This claim, however, was criticized largely on the basis of the methodological limitations of the research (Longstreth, 1981; Pinneau, 1955). Critics reported that much of the early literature provided few details regarding either the conditions in orphanages or the assessments used to evaluate children. Very often the number of children who were tested, the ages at which they were tested, and how often assessments were carried out were unspecified. Such limitations have made it difficult for current researchers to have confidence in the early data on institutionalized children. Not all early studies, however, predicted such dire outcomes for institutionalized children. In many parts of the world researchers K. MacLean were conducting studies in institutions attempting to ascertain the kinds of interventions that might prevent poor developmental outcomes. Such interventions included providing sensory stimulation (Broussard & Decarie, 1971), placing infants as “houseguests” with older residents (Skodak & Skeels, 1945, 1949), and improving child to caregiver ratios (Hunt, Mohandessi, Ghodessi, & Akiyama, 1976). All of these studies showed that simple changes within the orphanage environment increased children’s developmental competence. Further support for the idea that institutionalized children were not destined for developmental compromise came from the work of Barbara Tizard and her colleagues with children who had spent the first 2 years of their lives in high-quality institutions in the United Kingdom (Tizard, 1977). In these institutions the child to caregiver ratios were 3:1 and the children experienced adequate social stimulation, were taken on outings, and were fed well. The major way in which orphanage children’s lives differed from the lives of home-reared children was that caregivers were discouraged from forming intimate relationships with them (Tizard & Tizard, 1971). Tizard and Joseph (1970) first assessed children in the institution when they were 2 years old and compared them to a sample of home-reared children from a working class background. They found that the institution children’s IQ scores were only slightly lower than those of the working class children and that their language was only slightly delayed. At age 4.5 the mean IQ of the children in an institution was in the average range and no language problems were found (Tizard & Rees, 1974). Hodges and Tizard (1989) reported that children adopted from the institution had normal IQs at 8 and 16 years of age. Tizard’s findings were more positive than the findings of the earlier research, but it is important to note that the children in Tizard’s sample had not experienced the extreme deprivation experienced by earlier samples of children (e.g., Goldfarb, 1945a). Therefore, Tizard’s more positive findings may be partially explained by less severe deprivation. Although this research challenged Spitz’s claim that the damage resulting from institu- Institutionalization and child development 857 tionalization was irreparable, it did not ad- velopment (Bayley, 1969), Kaler and Freedress the issue of the extent to which length man (1994) found that none of the orphanage of institutionalization made a difference in de- children were functioning at their age level, velopmental outcomes. Dennis (1973) followed and 20 of the 25 children functioned at levels children from a Lebanese orphanage after they that were less than half their chronological were adopted and compared the developmen- age. Children’s Bayley scores were unrelated tal outcomes of children adopted at different to their Apgar scores at birth, suggesting that ages. He claimed that children who had been these data could not be explained by any iniadopted before 2 years of age eventually re- tial biological insult. Similarly, Carlson and gained normal IQs whereas those who were Earls (1997) reported that a group of 2- to 9adopted after 2 years of age showed perma- month-old infants in a Romanian orphanage nent deficits in IQ. Although this claim has scored well below the Bayley norms for their often been cited, it has not been adequately age. Further support for these findings was tested. provided by Sloutsky (1997), who assessed Given that much of the data from this ear- differences in IQ scores between 6- to 7-yearlier research was descriptive, anecdotal, and old children reared in a Russian orphanage short term, making it difficult to evaluate, and children of the same age reared at home more recently researchers have tried to ex- and found that the orphanage children scored amine larger samples of children using stan- lower in IQ than home-reared children. dardized measures of DQ, IQ, and academic Other researchers have been able to evaluachievement. Although their findings are more ate the intellectual development of previously hopeful, they are not inconsistent with the institutionalized Romanian children after the early literature. Both sets of literature clearly profound intervention of adoption. These two demonstrate that institutionalization early in projects, which are ongoing and have now follife has a negative impact on intellectual de- lowed children up to 8 or more years after velopment and that it is not only institutional- adoption, have provided the clearest informaization but also the length of institutionaliza- tion we have regarding intellectual develoption that is important. ment in postinstitutionalized children. The first Most of the recent information we have project, in British Columbia, Canada, comconcerning the impact of institutionalization prised an initial sample of 46 Romanian chilon intellectual development has been the re- dren who had spent at least 8 months (range = sult of the 1989 overthrow of the Ceausescu 8–53 months) in a Romanian orphanage (orregime in Romania, after which the world be- phanage group) prior to their adoption to Cancame aware of thousands of children being ada, when they were a median 18.5 months housed in Romanian state-run orphanages. De- old (range = 8–68 months); 29 children who scriptions of these orphanages have been re- would have gone to orphanage in Romania ported elsewhere (Fisher, Ames, Chisholm, & had they not been adopted prior to 4 months Savoie, 1997; Groza & Ileana, 1996), and it of age (early-adopted group); and 46 Canais clear that they were as bad as or worse than dian-born, nonadopted, never institutionalized the conditions reported in the earlier litera- children (Canadian-born group) who were inture. dividually matched on demographic characShortly after the revolution, Kaler and teristics to children in the orphanage group Freeman (1994) were able to assess children (Ames, 1997). The second project (Rutter & within the orphanage context. They conducted the ERA Study Team, 1998) comprised a samcognitive assessments on 25 children ranging ple of 165 children adopted to the United in age from 23 to 50 months who resided in Kingdom from Romania between birth and 42 a Romanian orphanage and compared them to months of age. Most of the children, but not same age peers from a Romanian kindergarten all, had been adopted from orphanages. These class. Apgar scores that were available for 13 children were compared to 52 children who of the children indicated that they had normal were adopted within the United Kingdom bebirths. Using the Bayley Scales of Infant De- fore 6 months of age. 858 Morison and colleagues (Morison, Ames, & Chisholm, 1995; Morison & Ellwood, 2000) examined development in the Canadian sample of Romanian orphanage children, once based on retrospective reports from when their parents first met them, once when the children had been in their adoptive homes for 11 months, and once when they had been in their adoptive homes for approximately 3 years. Based on parents’ reports of their children’s developmental condition when they first met them, Morison et al. (1995) found that 78% of orphanage children were delayed in all four areas of development (personal–social, gross motor, fine motor–adaptive, and language development) assessed by the Revised Denver Prescreening Developmental Questionnaire (RDPDQ; Frankenberg, 1986). There was not a particular area of development in which orphanage children were specifically compromised; rather, delay was pervasive across all areas. Rutter and his colleagues reported similar findings with their sample of adoptees in the United Kingdom (Rutter & the ERA Study Team, 1998). By 11 months postadoption, improvement was evident in the Canadian sample (Morison et al., 1995). At that time the majority of orphanage children remained delayed in two or more areas of development according to their parents’ report on the R-DPDQ. In addition, Revised Gesell Developmental Schedules (Knobloch, Stevens, & Malone, 1980) administered to 23 of the orphanage children showed that, although children were progressing at more than 1 month developmentally for each chronological month in Canada, their developmental quotients in areas of gross motor, adaptive, personal–social, and language averaged in the borderline range (68–85) and their fine motor abilities averaged in the low end of the average range (85+). Morison and colleagues (Morison et al., 1995) examined whether there were particular correlates associated with children’s intellectual test scores at 11 months postadoption. Although no demographic characteristics of the family (i.e., parents’ age and education, family income, and socioeconomic status) were associated with development at this time, there were characteristics of children’s institutional experience K. MacLean that were related to test scores. The length of time that children spent in the orphanage was positively associated with the number of areas of delay on the R-DPDQ and negatively correlated with children’s scores on the adaptive, personal–social, and language scales of the Gesell. In addition to the length of time in an institution, the availability of toys and having been a favorite in the institution were associated with fewer delays and higher scores on the Gesell scales, whereas being described as dirty when first met by parents was associated with more delays and lower Gesell scores (Morison et al., 1995). Three years postadoption these researchers (Morison & Ellwood, 2000) evaluated children’s intellectual development using the Stanford–Binet Intelligence Scale, Fourth Edition (SB4: Thorndike, Hagen, & Sattler, 1986). Given Dennis’ (1973) earlier claim, IQ results were analyzed separately for orphanage children who had been adopted before and after they were 2 years old. In comparing IQ scores among the younger children, who were 4.5 years old when assessed, Morison and Ellwood (2000) found a clear ordering among the groups, in which the Canadian-born group scored highest (M = 109), the early-adopted children scored in the middle (M = 99), and the orphanage children scored lower (M = 91). The orphanage children adopted at later ages (24–60 months old) had the lowest IQs of all (M = 68). Le Mare, Vaughan, Warford, and Fernyhough (2001) conducted a later follow-up of the Canadian sample, when the children were 9.5 years old or older. The pattern of group results remained consistent with the earlier findings. Canadian-born children, whose families were matched to the other groups on demographic variables, scored highest (M = 108), early-adopted children scored lower (M = 99), orphanage children adopted between 8 and 24 months of age scored even lower (M = 89), and orphanage children who had been adopted after 2 years of age scored lowest of all (M = 71). Not unlike their earlier performance on the R-DPDQ, the scores of later-adopted children were lower across all scales of the SB4 (i.e., overall IQ, verbal comprehension, and nonverbal reasoning) at both 3 years after Institutionalization and child development 859 adoption (Morison & Ellwood, 2000) and 8 age scored at the low end of the normal range years after adoption (Le Mare et al., 2001), of intelligence (M = 90). This difference may demonstrating that orphanage experience had be explained by considering differences in a general impact on all areas of intelligence. group composition between the two studies. Rutter and colleagues used the General The orphanage children in the Canadian study Cognitive Index of the McCarthy Scales (Mc- had all spent at least 8 months in an orphanCarthy, 1972) and reported a similar ordering age and had spent the great majority of their among their groups of 4-year-old (Rutter & lives in institutions. The correlation between the ERA Study Team, 1998) and 6-year-old age at adoption and total time in institution children (O’Connor, Rutter, Beckett, Keave- was +.97 (Morison & Ellwood, 2001). In conney, Kreppner, & the ERA Study Team, 2000), trast, close to 15% of the U.K. sample did not with within-U.K. adoptees and Romanian chil- experience an institutional upbringing: some dren adopted before 6 months of age scoring children had been reared in a family setting better than Romanian children adopted be- and others had been institutionalized for as littween 6 and 24 months of age, who in turn tle as 2 weeks (Rutter & the ERA Study Team, scored better than Romanian children adopted 1998). Inclusion of children with little or no between 24 and 42 months of age. This con- orphanage experience may have contributed sistent ordering of groups across studies dem- to higher average IQ scores in the children onstrates the negative and cumulative effect adopted after age 2 to the United Kingdom of institutionalization on IQ. Institutionaliza- than in their counterparts in the Canadian tion may not, however, be the only factor un- sample. derlying this ordering. The fact that the IQs Poor academic achievement has been found of the early-adopted children in the Canadian in both children in orphanages and children study were lower than those of Canadian-born after adoption from orphanage. A study of children in homes matched on demographic children reared in Greek orphanages in which variables also suggests possible effects of ge- conditions were far superior to the conditions netic background, prenatal and perinatal care, found in Romania found lower academic perand adoption, which were shared by the early- formance among 9-year-old orphanage chiladopted and orphanage children (Morison & dren compared to same gender and age peers Ellwood, 2000). living at home (Vorria, Rutter, Pickles, WolThe importance of length of institutional- kind, & Hobsbaum, 1998). Le Mare et al. ization for IQ is also supported by several re- (2001) studied children after they had been ports of negative correlations between the two in Canada 8 years or more and found that, variables: r = −.75 3 years after adoption according to teachers’ reports of academic (Morison & Ellwood, 2000), r = −.48 at age performance and results on a standardized 6 years (O’Connor, Rutter, Beckett, et al., achievement test, the Canada Quick Individ2000), and r = −.44 at 8 or more years after ual Educational Test (Wormelli & Carter, 1990), adoption (Le Mare et al., 2001). Several re- Canadian-born children performed best, orports (Morison & Ellwood, 2001; Rutter & phanage children adopted before 2 years of the ERA Study Team, 1998; O’Connor, Rut- age and early-adopted children (adopted beter, Beckett, et al., 2000) found that length of fore 4 months) obtained intermediate scores, institutionalization was the best predictor of and orphanage children adopted after 2 years children’s IQ when entered in multiple regres- of age performed the worst. In addition, 12% sions with other variables. of orphanage children adopted before 2 years Compared to the Canadian sample (Mori- of age and 60% of orphanage children adoptson & Ellwood, 2000) in which orphanage ed after 2 years of age had repeated a grade. children adopted after 2 years of age scored In comparison, only one early-adopted child in the low end of the “slow learner” range on and no Canadian-born children repeated a IQ (M = 68), the Romanian children in the grade in school. Le Mare et al. (2001) conU.K. sample (O’Connor, Rutter, Beckett, et cluded, however, that even though many oral., 2000) who were adopted after 2 years of phanage children were struggling within the 860 school context, a noticeable number of them were functioning well within the average range academically. There is no doubt that the intervention of adoption out of orphanage is a powerful one; this is supported by the recent literature demonstrating improvements in IQ after adoption (Le Mare et al., 2001; Morison et al., 1997; Morison & Ellwood, 2000; Rutter & the ERA Study Team, 1998; O’Connor, Rutter, Beckett, et al., 2000). Undoubtedly, every child is doing better intellectually than he or she would be doing had they remained in an orphanage, but the environment they encounter postadoption also makes a difference. Morison and Ellwood (2000) found that not only institutionalization but also the home environment contributed to children’s IQ scores 3 years after adoption. Children’s home environments were measured using the Home Observation for Measurement of the Environment (HOME; Caldwell & Bradley, 1984). The more stimulating and supportive the home environments provided by parents, the higher the IQs of the children. Morison and Ellwood (2000) found in a multiple regression that in addition to length of institutionalization and children’s R-DPDQ scores when their parents first met them, their scores on the HOME contributed significantly to intellectual outcomes. Development of children in institutions and children adopted from institutions has been studied for over 60 years. The strongest finding from the many studies is that spending time in an orphanage is related to lower DQs, IQs, and academic achievement. The longer the length of institutionalization, the greater the decline in these measures. After removal from orphanage, children improve on developmental and intellectual measures, but those who have been institutionalized for a long period of time may still show significant delays for many years after adoption. The effect of length of institutionalization on development has been demonstrated both through correlations and through group differences, most often by comparing groups that spent less than 2 years with those who have spent more than 2 years in an orphanage. The division at 2 years arose for two reasons: first, because of K. MacLean Dennis’ (1973) assertion (based on insufficient evidence) that children adopted before 2 years of age eventually regained normal IQs whereas those who were adopted after age 2 showed permanent deficits, and second, because the great majority of children adopted from orphanages since 1990 have been adopted before the age of 2, thus making it difficult to get an “older-adopted” group if group division lines are set higher than the 2-year mark. The findings obtained with the correlational and the group differences data are congruent, but it is unfortunate that under-2 versus over2 comparisons have become so common. Inspection of scatterplots of correlational data reveals that the relationship of length of institutionalization to cognitive measures is generally continuous and does not show any clear dividing point at 2 years of institutionalization. The older children are when adopted, the more difficult is the task, because they not only have greater deficiencies to make up but also have less time to recover before they have to cope with formal school classes. Because of this, many of them repeat a grade in school. There are, however, some orphanage children who in spite of their poor start are functioning well academically and appear not to have been greatly intellectually compromised by their early experience. Another common finding is that both conditions in orphanages and conditions in adoptive homes can modify the amount of cognitive deficit. Children do better if they are reared in better quality institutions (e.g., those studied by Tizard, 1977) or in institutions in which perceptual or social environments have been enriched. They also have higher IQs and academic achievement after adoption if their home environments are more stimulating and supportive. Although comparisons of previously institutionalized children with home-reared children suggest that institutionalization per se is the operative factor affecting their performance, the reason that children end up in orphanages must not be forgotten. Whereas there are cases in which children from sound biological, economic, and social backgrounds are institutionalized, it is more commonly the case that chil- Institutionalization and child development dren require institutional care because of a parental problem that may have genetic or behavioral implications for the child (e.g., alcoholism, drug abuse, mental retardation, mental illness) or because of social or economic conditions in the family that also mean that children receive poor prenatal or perinatal care or even abuse before entering an orphanage. The possible effect of these extrainstitutional, but associated, factors is shown when children with little or no institutional experience but from the same backgrounds as institutionalized children (e.g., the early-adopted group in Morison et al., 1995) score lower than home-reared comparison groups matched on demographic characteristics of their homes. Physical Development Dana Johnson and colleagues (Johnson et al., 1992) examined the medical condition of 65 previously institutionalized Romanian children who were adopted to the United States. They found that only 15% of these children were considered physically healthy at the time of adoption. Convergent evidence for this finding comes from the Romanian children adopted to Canada (Fisher et al., 1997). In this sample, 85% of children had a reported medical problem. Similarly, Hostetter, Iverson, Thomas, McKenzie, Dole, and Johnson (1991) found that at least one medical problem was reported for a majority of children adopted from 15 countries in East and South Asia, Central and South America, Africa, and the Caribbean and Pacific Islands. Fisher et al. (1997) found that according to their parents’ reports orphanage children’s most common medical problems when they were first adopted were intestinal parasites (31% of children), hepatitis B (28% of children), and anemia (15% of children). These medical problems are consistent with those noted by other researchers who have examined medical problems in previously institutionalized children from Romania (Benoit, Jocelyn, Moddeman, & Embree, 1996; Gyorkos & MacLean, 1992; Jenista, 1992; Marcovitch, Cesaroni, Roberts, & Swanson, 1995) and from several other countries (Hoksbergen, 1981; Hostetter et al. 1991). Hoksbergen surveyed parents from The Netherlands 861 who had adopted children from Korea, Bangladesh, Columbia, India, and Indonesia and found that malnutrition, intestinal parasites, and skin diseases were common. Fisher et al. (1997) found that 3 years after adoption the Romanian children’s medical problems had greatly improved but that children from Romanian orphanages still had more medical problems than both the early-adopted and Canadian-born comparison groups. In addition to having particular medical problems, previously institutionalized children are typically small and malnourished after the orphanage experience. Eighty-five percent of orphanage children in the Canadian sample fell below the 10th percentile and 59% of children fell below the 5th percentile for weight (Morison et al., 1995). In another group of 16 Romanian children adopted to Canada, 50% were below the 5th percentile for weight and 44% were below the 5th percentile for height (Benoit et al., 1996). Similarly, in the U.K. sample of Romanian adoptees, Rutter and the ERA Study Team (1998) reported that half of the children had heights, weights, and head circumferences below the 3 percentile. The evidence of delays in physical growth among previously institutionalized children is fairly consistent, and such delays are still apparent up to 3 years postadoption (Carlson & Earls, 1997; Chugani et al., 2001; Johnson et al., 1992; Rutter & the ERA Study Team, 1998). In several studies growth retardation in height has been shown to be associated with the length of time that children had spent in institutions (Ames, 1997; Johnson et al., 1992; Rutter & the ERA Study Team, 1998). The longer that children spent in an orphanage, the shorter they were for their age. Very few studies of orphanage-reared children have focused on the impact of institutionalization on physiological development. The only physiological measures that have been examined are cortisol levels (Carlson & Earls, 1997; Gunnar et al., 2001) and patterns of brain glucose metabolism (Chugani et al., 2001). Several researchers have proposed that the impact of early adversity on physical and psychological development is mediated, in part, through effects on the stress-sensitive hypothalamic–pituitary–adrenocorticol (HPA) sys- 862 tem (Gunnar, 2000; Heim, Owen, Plotsky, & Nemeroff, 1997). This hypothesis is based on substantial preclinical evidence that disturbances in caregiving early in life alters development of the HPA axis and its central releasing hormone (Sanchez, Ladd, & Plotsky, 2001). Although studies in rodents suggest that early deprivation of maternal care produces hyperresponsivity of the HPA system, studies in nonhuman primates have yielded evidence of disturbances in the HPA diurnal rhythm and low, rather than elevated, basal levels of cortisol, the hormonal product of this system (Dettling, Feldon, & Pryce, 2002). Both hyper- and hypocortisolemia can have negative effects on health and behavioral functioning (McEwen, 1998), although the impact of either type of dysregulation during periods of rapid brain development is not well understood. The first study to examine cortisol levels among orphanage children was conducted by Carlson and Earls (1997). They measured ambulatory cortisol levels in 2-year-old children residing in a Romanian orphanage and compared them to a sample of 2-year-old familyreared children. Carlson and Earls found that orphanage children’s cortisol levels were not elevated as compared to family-reared children. Group differences were apparent, however, in the pattern of cortisol production over the day. Family-reared children displayed the typical pattern: cortisol levels were highest in the early morning and decreased over the day. Orphanage children, on the other hand, did not display this pattern; indeed none of the children they studied exhibited the typical diurnal pattern of cortisol production. Noon levels of cortisol were positively correlated with delays in cognitive functioning in the orphanage group. Carlson and Earls (1997) suggested that orphanage experience disturbs the diurnal pattern of cortisol production but does not result in increased levels of cortisol concentrations overall. Gunnar and colleagues (2001) examined cortisol levels in the Canadian sample of Romanian adoptees tested 6.5 years after adoption to determine whether adoption into families would normalize the diurnal activity of this axis. Children who had spent more than 8 K. MacLean months in Romanian orphanages prior to their adoption were compared to children adopted from orphanage before they were 4 months of age (early-adopted group) and to Canadianborn children. The children, who ranged in age from 6 to 12 years at the time of cortisol sampling, all displayed the normal diurnal rhythm in cortisol. The orphanage children, however, had higher cortisol levels than children in the comparison groups. Evening levels were most highly correlated with the length of time children had spent in institution. Children’s cortisol levels were unrelated to either IQ scores obtained several years prior to cortisol assessment or their age at cortisol measurement. This was a small sample of children (n = 16 orphanage group children), and thus the generalizability of these results to other institutionalized children must be viewed with caution. Chugani and colleagues (2001) examined brain dysfunction in ten 7- to 11-year-old children who had spent on average 38 months (range, 16–90 months) in a Romanian orphanage prior to their adoption to the United States. At the time of assessment, children had been in their adoptive homes for an average of 67 months (range = 15–113 months). Abnormalities in brain glucose metabolism were examined using functional neuroimaging with positron emission tomography. The orphanage children were compared to 17 normal adults and to 7 children (age range = 7–13 years) with medically refractory focal epilepsy but who had normal brain glucose metabolism in the unaffected hemisphere. Chugani et al. (2001) found lower glucose metabolism in several areas of the brain among Romanian adoptees than in either the adult or child comparison groups. They suggested that the decreased brain glucose metabolism likely resulted from the Romanian adoptees’ experience of early deprivation and may be one explanation for the cognitive and behavioral deficits seen in many previously institutionalized children. The reader is cautioned that early deprivation in Chugani et al’s sample appears to have included an unusually high frequency of physical abuse. Chugani et al. (2001) reported that 5 of the 10 Romanian children they assessed had physical scarring that ap- Institutionalization and child development peared to have resulted from lacerations, burns, or broken bones. This rate of physical abuse is much higher than in any other study of Romanian adoptees. It is possible therefore that the brain dysfunction found by Chugani et al. (2001) may be at least partially related to pain and fear beyond the levels typical of institutional life. In summary, studies agree that children adopted from orphanages tend to be in poor health, malnourished, and small. The longer children remain in an orphanage, the shorter they are for their age. There is some evidence that orphanage children lack the normal diurnal pattern of cortisol production and have lower than normal brain glucose metabolism, but the full meaning of these findings is unclear. Further studies using a wider range of physiological variables will be necessary to elucidate this area of research. Behavior Problems Across studies researchers have consistently found that children with orphanage experience have more behavior problems than children without orphanage experience. Goldfarb (1943b) found that previously institutionalized 6- to 8-year-old children who had been placed in foster care at age 3 scored higher on behavior problem checklists than noninstitutionalized children of the same age who had spent their lives in foster care. The same difference between these groups was also found during adolescence (Goldfarb, 1943a). Even children who were adopted from the relatively benign orphanage environment studied by Tizard were described by their teachers as having more behavior problems than their classmates at both 8 (Tizard, 1977) and 16 years old (Hodges & Tizard, 1989). Verhulst, Althaus, and Versluis–Den Bieman (1990a, 1990b) found that 10- to 15-year-old children who had been adopted from other countries to the Netherlands between birth and 10 years of age had more behavior problems than sameage nonadopted children. Recent studies of Romanian adoptees have found similar results but have gone further in specifying the problems and tracking their changes over time. Fisher et al. (1997) found 863 that 11 months after adoption Romanian orphanage children adopted to Canada had higher internalizing and total scores on the Child Behavior Checklist (CBCL; Achenbach, Edelbrock, & Howell, 1987) than did early-adopted or Canadian-born children. No differences were found between the groups on externalizing behavior problems at that time. When children had been in Canada for 3 years, however, their externalizing and total scores were higher than those of Canadian-born children (Ames, 1997), and these scores remained higher than scores in the comparison groups more than 8 years after adoption (Warford, 2002). Marcovitch et al. (1997) reported similar results in their sample of 3- to 5-year-old Romanian children. Children adopted after 6 months of age had higher externalizing and total scores on the CBCL than Romanian children adopted before 6 months of age. In the Canadian sample (Ames, 1997) it initially appeared that in addition to becoming more externalizing, orphanage children had become less internalizing from 11 months to 3 years postadoption. However, this resulted from a difference in the items comprising the internalizing scales on the CBCL version for 2- to 3-year-olds used at the earlier time and the CBCL version for 4- to 18-year-olds used at the later time. Although 3 years after adoption the orphanage children no longer scored higher than comparison groups on the internalizing scale of the CBCL, they continued to score higher on the same internalizing items on which they had scored higher at 11 months postadoption, that is, stares into space, strange behavior, acts too young, and speech problems. After 3 years in Canada there was little change in their early internalizing behaviors. Indications that length of institutionalization is positively related to number of behavior problems have been found by Ames (1997), Beckett, Bredenkamp, Castle, Groothues, O’Connor, Rutter, and the ERA Study Team (2002), Fisher et al. (1997), Marcovitch et al. (1997), Verhulst et al. (1990a, 1990b), and Warford (2002). Having experienced institutionalization in the first year or two of life has been associated not only with children experiencing more behavior problems than normative samples 864 but also with particular kinds of behavior problems. In the Canadian sample, researchers conducted in-depth interviews with parents regarding any behavior problems they were experiencing with their children at or soon after adoption (Fisher et al., 1997). The main areas of problematic behavior were eating, stereotyped behavior, and peer relationships. Eleven months after adoption 65% of orphanage children had eating problems: parents reported that one-third of children refused to eat solid foods and almost one-third of children overate. The refusal to eat solid foods has also been reported for toddlers adopted from Bulgarian orphanages (Ripley, 1992) and children adopted to the United Kingdom from Romanian orphanages (Beckett et al., 2002). Fisher et al. (1997) suggested that these problem behaviors were directly attributable to orphanage life. It is not surprising that Romanian orphanage children had difficulty with solid foods, given that they were fed entirely by bottle for the first 1.5–2 years of life. Such behavior is not unlike the refusal to eat solid foods found among North American infants when the foods are first introduced (Wasserman, 1987). Parents in the Canadian study reported that during the first 11 months after adoption 85% of the orphanage children’s eating problems had improved or were completely resolved (Fisher et al., 1997). Beckett et al. (2002), however, reported that 15% of children adopted from Romania to the United Kingdom were still experiencing difficulties eating solid foods at 6 years of age. It appears this is a longstanding problem for some children. The problem of overeating was also reported in the earlier literature on institutionalization (Flint, 1978; Goldfarb, 1943b; Hoksbergen, 1981; Lowrey, 1940; Provence & Lipton, 1962; Tizard, 1977). In the Canadian sample, many parents of orphanage children reported that they continued to offer food to children after a meal and that it was never refused. Given orphanage children’s malnutrition and small size, such overeating probably represents their bodies’ natural catch-up mechanism. Children in orphanage had never eaten enough food to allow them to learn feelings of satiety (Fisher et al., 1997). Eating prob- K. MacLean lems have been found to decrease over time (Ames, 1997, Marcovitch et al., 1997). Three years after adoption, orphanage children had no more eating problems than did Canadianborn or early-adopted children (Ames, 1997). A common behavior problem found in most samples of orphanage children is stereotyped behavior, that is, repeating the same movement over and over, as in body rocking. In the Fisher et al. (1997) sample 84% of orphanage children displayed one or more stereotyped behaviors whereas no Canadian-born and only one early-adopted child had a stereotypy. Sixty-seven percent of children rocked, and 19% moved their hands stereotypically. In the U.K. sample, 47% of the Romanian children rocked when first adopted (Beckett et al., 2002). Marcovitch et al. (1995) and Groza and Ileana (1996) also reported stereotyped behaviors in their samples of Romanian children, and this behavior was previously described by Flint (1978), Goldfarb (1945a), and Tizard (1977). Fisher et al. (1997) suggested that the stereotyped behavior of orphanage children may represent the prolongation of the stereotyped behaviors found in normal infants around the time they learn to sit, stand, or walk (Thelan, 1979, 1981). It may also reflect a self-soothing strategy or an attempt to self-stimulate in the barren orphanage rooms. In the Canadian sample, stereotyped behavior was the most frequently reported behavior problem of orphanage children but it was also a behavior that improved quite rapidly (Fisher et al., 1997). Only 2% of orphanage children had not improved at all after 11 months in Canada. In some children, however, vestiges of stereotyped behavior persisted: 41% of orphanage children still displayed at least some of this behavior 3 years after adoption, although its frequency had decreased in all children (Ames, 1997). Beckett et al. (2002) also reported a decrease in stereotypies, but 18% of Romanian children in their U.K. sample still rocked at 6 years of age. The degree of improvement was negatively correlated with length of institutionalization (Fisher et al., 1997; Beckett et al., 2002). Investigators have also examined the question of whether orphanage children experi- Institutionalization and child development 865 ence more problems with peers than compari- from orphanage to foster care at age 3. Chuson groups of children. Based on parents’ gani et al. (2001) found that ten 7- to 11-yearreports when their children had been in Can- old children adopted from Romanian orphanada for 11 months, Fisher et al. (1997) found ages scored on standardized tests in the mildly that 32% of orphanage children had problems impaired range on attention and in the sewith peers. The most common problems in- verely impaired range on impulsivity. volved either avoiding contact with peers or At 3 years postadoption, orphanage children being overwhelmed by peers’ attention, neither in the Canadian sample scored higher than of which was reported for either Canadian- comparison group children on the Attention born or early-adopted children. Based on both Problems subscale of the CBCL (Ames, 1997). parents’ and teachers’ reports at 3 years after They were also reported by their parents acadoption, orphanage children continued to dis- cording to the Parenting Stress Index (Abidin, play more social behavior problems than the 1990) to be more distractible but not more hyCanadian-born children (Ames, 1997). Early- peractive on the Distractibility/Hyperactivity adopted children’s scores were intermediate subscale than comparison group children at between the other two groups. By the time the 11 months after adoption (Mainemer, Gilman, orphanage children had been in Canada for 8 & Ames, 1998) and more distractible and hyor more years, their peer relationships had im- peractive than comparison children 3 years afproved (Warford, 2002). Orphanage children ter adoption (Ames, 1997). Further support were not different from early-adopted or Ca- for these findings comes from the U.K. study, nadian-born children on acceptance ratings from in which Kreppner, O’Connor, and Rutter their school peers or on their parents’ ratings (2001) found that at both 4 and 6 years of of the number of good friends they had or the age the children who had been adopted from ease with which they made new friends. Their Romania after 6 months of age were rated by feelings of social competence and loneliness both teachers and parents on the Revised Rutwere also similar to those of the two compari- ter Parent and Teacher scales (Hogg, Rutson groups. On the other hand, orphanage ter, & Richman, 1997) as displaying higher children felt less social support from their inattention/overactivity than did Romanian peer group or a close friend than did children children or U.K. children who were adopted in the other groups. The longer children had before 6 months of age. spent in institution, the less peer support they The Canadian adoptees from Romania confelt. tinued to display more attention problems 8 or Another problem area concerns attentional more years after adoption (Le Mare & Audet, difficulties, which have been reported in chil- 2002): orphanage children scored higher than dren who reside in institutions and in most both Canadian-born and early-adopted children studies of previously institutionalized children. on the Attention subscale of the CBCL and Vorria, Rutter, Pickles, Wolkind, and Hobs- on the Impulsivity and Inhibitory Control subbaum (1998) found that, compared to same- scales of the Children’s Behavior Questionaged peers living at home, 9- to 11-year-old naire (Ahadi, Rothbart, & Ye, 1993) and lower children living in a Greek orphanage were more on the questionnaire’s Attention Focusing subinattentive and more often engaged in nonpro- scale. These findings were consistent across ductive activities in the classroom. Parents of both parent and teacher reports. Supporting previously institutionalized Romanian children these data was the additional finding that 29% who ranged in age from 7 months to 11 years of children in the orphanage group had re(median age = 3 years) reported high activity ceived a clinical diagnosis of attention-deficit levels, inability to attend, and distractibility as disorder (ADD) or attention-deficit/hyperacongoing problems in their children (Marco- tivity disorder (ADHD) whereas none of the vitch et al., 1995). These findings are consis- comparison group children had received such tent with the earlier work of Goldfarb (1945b) a diagnosis (Le Mare & Audet, 2002). who reported distractibility and a lack of conAcross studies, attentional difficulties have centration among adolescents who had moved been consistently related to the length of time 866 K. MacLean that children had spent in orphanage (Krepp- phanage. What causes some children to have ner et al., 2001; Le Mare & Audet, 2002; persistent rather than transient problems in Marcovitch et al. 1997). In addition, Le Mare these areas is still unknown. and Audet (2002) found that children’s attenPeer problems are also believed to arise tional difficulties at 8 years of age were nega- from the characteristics of the orphanage entively correlated with their HOME environ- vironment. Young children in Romanian orment scores (Caldwell & Bradley, 1984) at 3 phanages are generally quiet and unresponyears postadoption. The more nurturing and sive to each other, so after adoption they are stimulating the home environment, the fewer wary of and withdraw from family-reared chilattentional difficulties children experienced dren who are noisier, more active, and more later. It was not only institutionalization but unpredictable than their peers in the orphanalso the postadoption environment that aided age (Fisher et al., 1997). This unwillingness in the prediction of attention difficulties. to interact with peers further prevents them In accordance with early studies, the more from learning normal patterns of social interrecent studies of Romanian adoptees have action. When they do become comfortable shown that children reared in institutions have enough to interact with peers in their new enmore behavior problems than family-reared vironment, they externalize (“act out”) in unchildren, that the number of behavior prob- acceptable ways that do not endear them to lems is related to the length of institutional- those peers. Several years after adoption, they ization, and that some behavior problems last have improved on most indices of behavior up to 8 or more years after removal from the with peers but still have externalizing probinstitution. Fisher et al. (1997) have pointed lems and feel less social support from peers or out that most of the characteristic early prob- a close friend than family-reared children do. lems of children adopted from orphanages Even problems in attending may be at least might indicate brain damage or emotionally partly explained by the abrupt move from the produced pathology if they were found in very low levels of stimulation in the orphanchildren reared in families. In orphanage chil- age to the overwhelming sensory stimulation dren, however, they seem more correctly of an adoptive home without any opportunity characterized as behavioral adaptations to or- to gradually learn how to deal with greater phanage life. Children who have had inade- sensory complexity. Attention problems may quate nutrition in orphanage initially overeat go unnoticed when children are very young, until they reach the proper weight. Children but by preschool age they are apparent, and who have been fed entirely by bottle for the are often accompanied by hyperactivity. In first 2 years of life resist solid food when it is school years these problems may interfere introduced for the first time. Young orphan- with academic performance, and even lead to age children show the same stereotyped be- diagnoses of ADD or ADHD. haviors as family-reared children just starting to sit, stand, or walk (Thelen, 1979), but when Indiscriminate Friendliness restriction to a crib for 18–20 hr/day does not permit them to develop their motor skills Another behavior problem that is particular(Ames, 1997), these early stereotyped behav- ly enduring in previously institutionalized iors are practiced for a prolonged period. In children is indiscriminate friendliness. Tizard the general absence of sensory stimulation (1977) characterized indiscriminate friendliand soothing by adults, for example, by rock- ness as behavior that is affectionate and ing, children may learn to provide their own friendly toward all adults (including strangstimulation or soothing by making stereo- ers) without the fear or caution that is typical typed movements. Both eating problems and in young children. In these cases the chilstereotyped movements decrease with time dren’s behavior toward a stranger cannot be after adoption, but in each case there remains discriminated from their behavior toward their a small proportion of children who still have primary caregivers. It appears as though any the problem several years after leaving the or- adult is sufficient for the child as long as the Institutionalization and child development 867 child’s needs are met (Provence & Lipton, Further support for the suggestion that in1962). References to indiscriminately friendly discriminate friendliness is a characteristic bebehavior were evident in the early literature havior of orphanage children comes from the on the social development of institutionalized work of O’Connor and his colleagues in the children who were later fostered or adopted. United Kingdom (O’Connor, Bredenkamp, RutProvence and Lipton (1962) followed the ter, & the ERA Study Team, 1999; O’Connor, progress of 14 previously institutionalized Rutter, & the ERA Study Team, 2000). Using children who had been placed in foster care a scale of “disinhibited attachment disturbetween 18 and 24 months of age. They re- bance,” which contained three items similar ported that these children were indiscrimi- to those used by Chisholm (1998; Chisholm nately friendly to all adults. Tizard followed et al., 1995), O’Connor et al. (1999) found 24 children who had spent their first 2 years that approximately 20% of Romanian children in orphanage and were subsequently either adopted between 6 and 24 months of age had adopted or restored to their biological parents. high scores on the scale compared to approxiBased on parents’ reports these children dis- mately 10% of Romanian children adopted played indiscriminate friendliness to adults at before 6 months of age and 2% of a group of 4 (Tizard & Rees, 1975) and 8 years of age children adopted within the United Kingdom. (Tizard & Hodges, 1978), demonstrating the O’Connor et al. (2000a) found that high scores enduring nature of this behavior. A few adopt- were obtained by more 6-year-old Romanian ees were still indiscriminately friendly with children who had spent between 6 and 42 adults at 16 years of age (Hodges & Tizard, months in orphanage than by 6-year-olds who 1989). Goldfarb (1955) found that indiscrimi- had been adopted from either Romania or the nate friendliness was still present in adoles- United Kingdom before 6 months of age. cents who had been institutionalized as chilSeveral studies have examined indiscrimidren but who were later fostered to unstable nate friendliness over time. In the Canadian foster care placements. sample, orphanage children were just as indisMore recently, researchers have attempted criminate 3 years after adoption as they had to develop measures of indiscriminate friend- been 11 months after adoption, and 90% of liness. Chisholm (1998; Chisholm et al., 1995) parents reported no improvement in this bedeveloped a 5-item measure of indiscriminate havior over that period of time (Chisholm, friendliness. Parents were asked five questions 1998). In contrast, displays of indiscriminate assessing (a) whether their child wandered friendliness had decreased over the same time without distress, (b) whether their child was period in the early-adopted Romanian chilwilling to go home with a stranger, (c) how dren. In a follow-up study of the Canadian friendly their child was with new adults, (d) sample more than 8 years after adoption, Ferwhether their child was ever shy, and (e) what nyhough and colleagues (2002) found a positheir child typically did upon meeting new tive correlation between orphanage children’s adults. Orphanage children in the Canadian indiscriminate friendliness scores 3 years poststudy displayed significantly more indiscrimi- adoption and more than 8 years postadoption. nate friendliness than both early-adopted and There were no differences in children’s scores Canadian-born children at both 11 months at these two points in time. O’Connor and and 3 years after adoption (Chisholm, 1998). his colleagues reported a positive correlation Providing corroborative evidence for this find- between disinhibited attachment disturbance ing, 71% of parents in the orphanage group scores at 4 and 6 years of age for Romanian described their children as “overly friendly.” adoptees to the United Kingdom (O’Connor, Eight years after adoption, orphanage children Rutter, & the ERA Study Team, 2000). Sixtycontinued to display significantly more indis- two percent of children showed no category criminately friendly behavior than the Cana- change (“none,” “mild,” or “marked”) in these dian-born or early-adopted groups of children, behaviors from 4 to 6 years of age. who did not differ from each other (FernyUnlike the findings for intellectual develhough, Audet, & Le Mare, 2002). opment and for other behavior problems, there 868 K. MacLean is disagreement among studies as to whether children’s scores on indiscriminate friendliness there is a relationship between the length of at 3 years after adoption were positively cortime that children have spent in orphanages related with their attention problem scores more (or their age at adoption) and the amount of than 8 years after adoption. Consistent with indiscriminate friendliness they display. O’Con- the Canadian findings, in the U.K. sample nor and his colleagues found modest correla- children’s scores on disinhibited attachment tions at both 4 years (O’Connor, et al., 1999) disturbance were positively correlated with and 6 years of age (O’Connor, Rutter, et al., their scores on hyperactivity and disruptive 2000) between duration of deprivation (age at behavior measures at age 4 (O’Connor et al., adoption) and children’s scores on disinhib- 1999) and hyperactivity, disruptive behavior, ited attachment behaviors in the U.K. group. and emotional difficulties measures at 6 years They pointed out, however, that some chil- of age (O’Connor, Rutter, et al., 2000). dren who were adopted after 2 years in RomaIn contrast to behavior problems, IQ does nia showed no signs of disinhibited attach- not appear to be strongly related to indiscrimiment behaviors and other children who were nate friendliness. In the Canadian sample, adopted within the first 6 months of life showed Chisholm (1998) found no association between signs of the behavior (O’Connor, Rutter, et children’s IQ scores at 3 years postadoption al., 2000). Chisholm et al. (Chisholm, 1998; and their displays of indiscriminate friendliChisholm et al., 1995), Fernyhough et al. ness. Consistent with Chisholm’s finding, (2002), and Tizard et al. (Tizard & Hodges, O’Connor et al. (1999) found no correlation 1978; Tizard & Rees, 1975) all found no rela- between children’s IQ scores and disinhibited tionship between length of institutionalization attachment disturbance when the children in and indiscriminate friendliness. the U.K. sample were 4 years old. In their folIn an attempt to better understand indis- low-up study when children were 6 years of criminate friendliness, researchers have exam- age, O’Connor, Rutter, et al. (2000) reported ined particular correlates that may be associ- a significant correlation between disinhibited ated with this behavior, focusing on aspects attachment disturbance and IQ; but this relaof the institutional environment, child charac- tionship was no longer significant when the teristics, and family characteristics. In the Ca- analysis was controlled for age at adoption, nadian sample Chisholm (1998) found that suggesting that both IQ scores and disinhibhigh scores on indiscriminate friendliness 3 ited attachment behavior were the result of years postadoption were positively correlated duration of deprivation. with the child having been a favorite in the Researchers do not yet have a clear underinstitution. Other measures of the quality of standing of indiscriminate friendliness. Unlike institutions, for example, quality of physical many of the initial behaviors of concern in care, whether toys were available (Chisholm institutionalized children, displays of indiset al. 1998), or children’s weight or develop- criminate friendliness do not appear to dissimental delay at adoption (O’Connor et al., pate over time but are still in evidence up to 8 1999; O’Connor, Rutter, et al., 2000) were years after leaving the orphanage environment unrelated to children’s indiscriminate friendli- (Fernyhough et al., 2002; O’Connor, Rutter, ness. et al., 2000) and may persist even longer Particular characteristics of the child after (Goldfarb, 1955). Chisholm (1998) has sugadoption have also been examined as correlates gested that indiscriminate friendliness may of indiscriminate friendliness. Researchers have serve an adaptive function in the context of consistently found associations between or- the orphanage environment where emotional phanage children’s indiscriminate friendliness resources are extremely limited. Amid the and their behavior problems. Chisholm (1998) passivity of the orphanage an indiscriminately found that at 3 years postadoption, children’s friendly child may receive what little attention indiscriminate friendliness scores were posi- caregivers have to offer. The fact that indistively correlated with their scores on the criminate friendliness was associated with havCBCL. Le Mare and Audet (2002) found that ing been a favorite in the orphanage supports Institutionalization and child development 869 this contention but does not explain what were items that most clearly capture secure function such behavior serves after adoption. base behavior. Similarly, O’Connor et al. (2003) It may be behavior that is reinforced by both reported that children in their sample who parents and strangers. Chisholm et al. (1995) scored high on their Disinhibited Attachment reported that at 11 months postadoption, par- Disturbance scale were overwhelmingly clasents were pleased that their child appeared to sified as Insecure/Other in terms of their atbe fond of everyone. According to parents’ tachment patterns. Two of the three items on reports newly adopted orphanage children were O’Connor’s scale (readiness to go off with a often approached, talked to, and hugged by stranger, failure to check back with parent in total strangers, so it is not difficult to imagine new anxiety provoking situations) appear to that they felt that such behavior was appro- measure the same behaviors as Chisholm’s priate. two extreme items. Although orphanage chilGiven that indiscriminate friendliness in dren in the Canadian sample scored higher orphanage children has been consistently as- than comparison groups on each of the five sociated with attentional difficulties like dis- items on the indiscriminate friendliness meatractibility, lack of focus, impulsivity, and sure, the other three items on the scale (i.e., lack of inhibitory control, to some extent it being friendly with new adults, never having may reflect the expression of those difficulties been shy, and eagerly approaching new adults) as applied to interactions with strangers. On were not associated with children’s insecure the other hand, Chisholm et al. (1995) sug- attachment patterns and may measure simple gested that indiscriminately friendly behavior uninhibited friendliness rather than involveis not unlike the indiscriminate behavior seen ment of the attachment system. in infants prior to the formation of an attachBoth Chisholm (1998) and O’Connor et al. ment relationship and thus may reflect a de- (2003) found some securely attached children lay in the attachment system. O’Connor et al. who are indiscriminately friendly, and Mar(1999) also suggested that indiscriminate friend- covitch et al. (1997) reported that some seliness may represent a form of developmental curely attached children were observed to be exdelay in the attachment behavioral system, al- tremely friendly to the stranger in the Strange though their use of the term “disinhibited at- Situation. It seems unlikely, therefore, that intachment disturbance” seems to suggest atten- discriminate friendliness necessarily indicates tion problems as well as involvement of the an attachment disorder (Chisholm, 1998). attachment system. In summary, research has shown that indisIndiscriminate friendliness is particularly criminate friendliness is a common characterrelevant to the study of attachment, given sug- istic of previously institutionalized children, gestions that this behavior may be indicative which persists years after removal from orof “nonattachment” (Lieberman & Pawl, 1988) phanage. In contrast to the negative relationor the disinhibited/indiscriminate subtype of ships found between length of institutionalizareactive attachment disorder (Zeanah, 1996, tion and orphanage children’s height, IQ, 2000). Both Lieberman and Pawl (1988) and academic achievement, and behavior problems, Zeanah (2000) described these disorders as re- the relation of length of institutionalization to sulting from an infant not having had the op- indiscriminate friendliness has been found across portunity to form an attachment relationship. studies to be weak or nonexistent. This is precisely the situation of children reared Indiscriminate friendliness has little or no in orphanage environments, so researchers have relation to IQ, but it is positively correlated focused on linking this behavior to children’s with behavior problems, especially problems attachment. Chisholm (1998) found that the of attention. To some extent, therefore, it may two most extreme items on her measure of reflect distractibility and impulsivity as apindiscriminate friendliness (i.e., being willing plied to interactions with strange adults. It to go home with a stranger and wandering also has been theorized to represent a failure without distress) were associated with inse- or developmental delay in attachment. The cure attachment in the Canadian sample. These finding that only those indiscriminate friendli- 870 ness measurement items reflecting lack of secure base behavior, and not those reflecting eager approach and engagement with new adults, are related to insecure attachment suggests a possible differentiation of two components of indiscriminate friendliness: uninhibited friendliness toward all adults and lack of the secure base behavior characteristic of attachment. How distinguishable these two components are in terms of their antecedents and consequences remains to be explored further. K. MacLean vided one-on-one attention to children. Tizard and Rees (1975) described 4-year-old children’s behavior toward their caregivers as very clingy but claimed that the children did not care deeply about anyone. By the time the children were 8 years old there were only 7 institution children left to study and only a minority of them were suspected to have formed any attachment to their institutional caregiver. This work provides the only direct evidence that institutionalized children would have been unlikely to have the opportunity to form an attachment with their caregivers. Attachment Chisholm (1998) suggested that there are Researchers have focused enormous energy reasons to suppose why developing an attachon attempting to evaluate the impact of insti- ment relationship may be more difficult for tutionalization on the ability of children to previously institutionalized children. Accordform attachment relationships. Given that at- ing to attachment theory, a child’s attachment tachment usually develops some time during behavior becomes organized toward a particthe second half of the first year of life (Bowl- ular caregiver sometime between 6 and 12 by, 1969/1982) children who have been housed months of age (Bowlby, 1969/1982). An adein institutions during the first year or 2 of life quate caregiver readily responds to an infant’s necessarily develop an attachment relation- needs for close contact and understands an inship with adoptive parents later than is typi- fant’s distress if separated from the caregiver. cal. Although little evidence exists on this When a child is beyond 2 years of age, carepoint, most researchers have assumed that the givers may be less responsive to the child’s children would have been unlikely to have de- needs for close contact and may expect him veloped an attachment relationship with care- or her to display more autonomy. The child’s givers within the institution, given the very need for contact may be viewed as “clingy” high child to caregiver ratios within institu- rather than “cuddly” behavior, and a parent tions (Chisholm, 1998). In Romanian orphan- may not be as patient with such behavior. ages the child to caregiver ratios ranged from Given that attachment theory suggests that 10 to 1 for children under 2 years of age to as sensitive responsiveness predicts the quality high as 20 to 1 for children over 3 years of of the attachment relationship (Ainsworth, age. It is unlikely that caregivers in such a Blehar, Waters, & Wall, 1978; DeWolff & context would have had time to spend in sen- van IJzendoorn, 1997), if a caregiver is less sitive responsive care with particular children responsive, developing an attachment rela(Chisholm, 1998). tionship later may be more difficult. Chisholm Tizard has been the only researcher who et al. (1995) also reported that children from examined children’s behavior toward their institutions did not initially display proximitycaregivers within the institution context. Tiz- promoting behaviors like smiling, crying, and ard and Tizard (1971) found that when com- making eye contact, behaviors that often proparing children in institution to family-reared mote contact with caregivers. Parents in the 2-year-olds, a list of preferred persons could Canadian sample reported that when they first easily be constructed for family-reared chil- met their children fewer than half of the children whereas lists of preferred persons for dren would smile back at someone who smiled children from institutions included anyone at them. Based on parents’ reports at 11 months that children knew well. The only “favorite” postadoption, 14% of children did not show people in children’s lives were people they signs of experiencing pain and 31% of chilsaw rarely (i.e., a parent who visited), but dren would not signal parents upon waking such persons were the only ones who pro- (Chisholm et al., 1995). The absence of such Institutionalization and child development 871 behavior may also make it difficult for parents ports whether children had truly been unable to know when and how to respond appropri- to form an initial attachment with caregivers ately to their children. A third reason why de- in the institution. Tizard and Rees (1975) reveloping an attachment may be difficult is be- ported that even though the institutions maincause of the neglect children experienced in tained a policy that discouraged caregivers from the institution. As a result they may have de- forming relationships with particular children, veloped expectations of others as untrustwor- there were some indications that children did thy, and this might promote difficult or passive have preferences among their regular caregivinteraction styles that would have a negative ers. Perhaps most important, there was a radiimpact on a parent’s ability to be sensitively cal difference between the two studies in terms responsive. of the stability of placement for the children Findings in the early literature were incon- postinstitutionalization. Children in Goldfarb’s sistent regarding whether institutionalized study very often went to unstable foster care children were capable of forming an attach- placements. By the time children were bement with their adoptive parents. Goldfarb tween 10 and 14 years of age they had experi(1943a) found that the majority of 10- to 14- enced care in three or four different foster year-olds who had been in orphanage for the homes (Goldfarb, 1943a). It is likely that havfirst 3 years of life and were later removed to ing experienced a series of broken attachment foster homes were described as withdrawn and relationships contributed to children’s emo“removed” with both family members and their tional coldness. Children in Tizard’s sample, caseworkers, and appeared unperturbed by ei- however, went to stable adoptive homes, evither threats of removal from the home or denced by the report that only 14% of adopchanges in foster care placements (Goldfarb, tive placements had broken down by the time 1945b). None of the comparison children who children were 16 years old (Hodges & Tizard, had spent their lives in foster care were de- 1989). scribed in this way. Goldfarb concluded that More recently, researchers have examined orphanage children were unable to develop at- attachment in previously institutionalized chiltachment relationships with their foster par- dren using standard separation reunion proceents and that the effects of institutionalization dures and relying on validated attachment were permanent. In contrast to Goldfarb’s work, coding systems (Chisholm, 1998; Marcovitch Tizard (1977) concluded that children could et al., 1997; O’Connor et al., 2003). In each become attached to parents after leaving an of these studies researchers used a separation institution. In her sample, 20 of 25 children reunion procedure with preschool-aged chilwere reported to have formed an attachment dren in which children first interacted with relationship with their adoptive parents within their mothers in play, stayed in the same room a year of leaving the institution and a majority with a stranger while their mother left, and of parents reported that their child was deeply then were reunited with their mothers. This attached to them. procedure was carried out in children’s homes Ames and Chisholm (2001) have suggested by both Chisholm (1998) and O’Connor et al. several reasons for these inconsistent findings. (2003), but Marcovitch et al. (1997) carried First, the conditions in the orphanages studied out the procedure in a hospital laboratory. In by Goldfarb were much worse than the condi- all studies children’s behavior during this protions in the Tizard study. Children in Gold- cedure was videotaped and then coded by farb’s sample were delayed in every area of trained coders for quality of attachment. Chisdevelopment. In contrast, children in Tizard’s holm (1998) used Crittenden’s (1992a) Presample were chosen for good health and had school Assessment of Attachment (PAA) to average IQs when they left the institution code preschool attachment patterns, whereas (Tizard, Cooperman, Joseph, & Tizard, 1972). both Marcovitch et al. (1997) and O’Connor These factors would bode well for the forma- et al. (2003) used the Cassidy, Marvin, and tion of an attachment relationship with par- MacArthur Working Group (1992) Preschool ents. Second, it is unclear from Tizard’s re- Attachment System. Both of these coding sys- 872 K. MacLean tems have reported validity (Solomon & Dijkstra, & Bus, 1995). In addition, in all George, 1999). three studies (Chisholm, 1998; Marcovitch et Marcovitch et al. (1997) examined attach- al., 1997; O’Connor et al., 2003), fully one ment in a sample of 4-year-old Romanian third of children were classified as securely children adopted to Canada who had spent 6 attached. This clearly argues against the Goldmonths or more in an institution (institution farb contention that children from institutions group) and compared them to Romanian chil- are unable to form an attachment relationship dren who had spent less than 6 months in in- with their adoptive parents. stitution (home group). A second comparison On the other hand, it is clearly the case group comprised healthy 4-year-old Canadian that across studies the percentage of securely children. O’Connor et al.’s (2003) sample attached children in any previously institucomprised two groups of 4-year-old children tionalized sample was lower than the percentadopted from Romania, one group adopted age of secure children in any comparison before 6 months of age and the other adopted sample (Chisholm, 1998; Marcovitch et al., between 6 and 24 months of age. Most, but 1997; O’Connor et al., 2003). The one excepnot all, of these children had been adopted tion to this general conclusion is that no diffrom orphanages. The comparison sample com- ferences were found in the Marcovitch study prised U.K. born children who were adopted between their institution and home groups. before 6 months of age. Chisholm examined Ames and Chisholm (2001) have suggested attachment in 30 orphanage children aged 54 that failure to find a clear difference between months and 13 orphanage children who ranged these groups may be due either to the small in age from 65 to 110 months. Each of these number of children in the sample who had children had spent between 8 and 53 months more than 6 months of institutionalization or in orphanage. These children were individu- to the mixed backgrounds of the home samally matched on age and sex to children in ple. The only criterion for inclusion in the the early-adopted (adopted before 4 months of home group was that children had spent less age) and Canadian-born comparison groups, than 6 months in either a hospital or orphanand groups did not differ on family demo- age in Romania. There may have been some graphic variables. children in the home group who had spent Across all three studies there was no evi- close to 6 months in orphanage and therefore dence of a child being unattached to his or her there would be little difference in the amount caregiver. It is important to note, however, of institutional experience between the two that the coding systems used were initially de- groups. signed to evaluate the quality of attachment In all three studies a large percentage of rather than the presence or absence of an at- previously institutionalized children had attachment relationship. Nevertheless, children’s tachments to their primary caregivers that attachment patterns were related to other mea- were either secure or typical insecure (i. e., sures in ways consistent with attachment the- insecurely attached in a manner that is comory, which provides some construct validation mon in normative North American samples). for use of these measures with institutional- In Chisholm’s (1998) study, 67% of the orized children (Chisholm, 1998; O’Connor et phanage children had either secure or typical al., 2003). Chisholm (1998) found that, com- insecure attachment patterns. In other words, pared to orphanage children classified as se- they had developed attachment patterns that cure, those who were classified as insecure were identical to those displayed by 95% of had lower IQs, more behavior problems, par- the Canadian-born and early-adopted children. ents who reported higher levels of parenting This finding suggests that in spite of having stress, and lower socioeconomic status back- spent 8 months or more in an extremely degrounds. Each of these variables has been as- priving environment and developing an atsociated with insecurity in previous studies of tachment later than is typical, most orphanage attachment (Greenberg, 1999; Teti, Gelfand, children were still able to form an attachment Messinger, & Isabella, 1995; van IJzendoorn, relationship with their adoptive parents that Institutionalization and child development 873 was similar to those found in normative sam- This is not the case for most previously instiples. tutionalized children: although orphanage chilIt was also the case in all three studies, dren have experienced extreme neglect, it was however, that a considerable number of or- not perpetrated by their postadoption attachphanage children had developed very unusual ment figures. and atypical insecure attachment patterns with Orphanage children in the Canadian samtheir adoptive parent, which were rare in com- ple also displayed a higher percentage of comparison group children. Chisholm (1998) re- pulsive caregiving (A3), a pattern more comported that 33% of her orphanage sample had monly seen in neglected samples than in abused developed an atypical insecure attachment pat- samples of children (Crittenden & Claussen, tern, as compared to 7% of Canadian-born and 2000). Twelve percent of orphanage children 4% of early-adopted children. In parallel to displayed a compulsive caregiving pattern, as Chisholm’s findings, both O’Connor et al. compared to 4% of early-adopted Romanian (2003) and Marcovitch et al. (1997) found that children and 2% of Canadian-born children. slightly over 40% of their orphanage samples The compulsive caregiving pattern is adopted had developed an atypical insecure attach- in response to a withdrawn or unresponsive ment pattern. Such atypical insecure patterns caregiver. A child displaying an A3 pattern are rare in normative samples and are found inhibits negative affect and attempts to cheer more commonly in clinical samples of mal- an unresponsive caregiver with overbrightness treated children (Carlson, Cicchetti, Barnett, and nurturance (Crittenden, 1992a). Critten& Braunwald, 1989; Cicchetti & Barnett, den suggested that in the case of neglect, such 1991; Crittenden, 1988; Lieberman & Zeanah, a strategy increases the likelihood of a child 1995). Some researchers have suggested that gaining the parents’ attention. Children’s exsuch attachment patterns may be risk factors perience of extreme neglect earlier in their in the development of psychopathology (Carl- lives may have resulted in their developing a son & Sroufe, 1995). compulsive caregiving strategy that would enThe distribution of atypical patterns in pre- sure they received what little attention was viously institutionalized children is not the available to them in the orphanage context. same as in other maltreated samples. One dif- For some children, this seems to continue in ference is that orphanage samples display a their adoptive homes, especially if a parent is lower percentage of the Defended/Coercive withdrawn because of the burden of dealing (A/C) insecure pattern than do other maltreated with the child. Chisholm (2000) described such samples. A child classified as A/C shifts back a case in which an overburdened parent was and forth between defended and coercive strat- withdrawn and unresponsive in interaction egies in response to the behavior of an often with her child. The child displayed false, overunpredictable caregiver. Twelve percent of bright affect and took most of the responsibilthe Canadian orphanage sample was classified ity for maintaining the interaction with the as A/C. Whereas this percentage is higher parent. This child was classified as compulsive than is found in normative samples of chil- caregiving (A3) and may have implemented dren, it is far lower than the 58% of A/C pat- such a strategy because of his experience of externs reported by Crittenden (1988) and the treme neglect in orphanage, experience with an 27% reported by Cicchetti and Barnett (1991) unresponsive caregiver, or both. for maltreated children who have come to the Another difference is that the Insecure attention of social services. The characteris- (other) pattern is more common in orphanage tics of the parents may explain this difference samples than in maltreated samples. Insecure between institutionalized samples and other (other), by definition indicates behavior that maltreated samples of children. In maltreated is difficult to classify. This classification is samples, children are subjected to abuse and/ given when a child is clearly insecure, but the or neglect at the hands of their attachment fig- strategy that he or she uses in interaction does ures and have developed a flexible, organized not fit any of the established insecure patstrategy for coping with an abusive caregiver. terns. Nine percent of the Canadian orphanage 874 K. MacLean sample was classified as Insecure (other), problems, parenting stress, and indiscriminate whereas Crittenden (1992a) only found 3 or 4 friendliness and higher on IQ. Chisholm (1996) out of 100 maltreated children classified as concluded that Secure (other) children were Insecure (other). more similar to children classified as secure Such atypicality in orphanage children’s than to children classified as Insecure (other). attachment was also apparent in the attach- This suggests that the PAA is able to classify ment patterns of orphanage children classified children appropriately as secure and insecure, as secure. The PAA (Crittenden, 1992a) has a and yet reveals that orphanage children may secure classification labeled “Secure (other).” display even secure patterns of attachment difThis classification is given when children are ferently from both normal samples and noninclearly secure but the strategies they use in stitutional, high-risk samples of children. For interaction with their caregiver do not reflect example, in a case study Chisholm (2000) deany of the standard subpatterns of secure. Crit- scribed the behavior of a previously institutenden has not reported any children classified tionalized child who had been classified as as Secure (other) in her published results (Crit- Secure (other). During a separation–reunion tenden, 1985, 1988, 1992b; Crittenden, Par- procedure the child’s interaction with her tridge, & Claussen, 1991) with maltreated mother was warm and relaxed, prior to sepsamples, and Teti and his colleagues (Teti et aration there were clear indicators of open al., 1995) reported only 1 child classified as negotiation regarding the separation, and neg“Secure (other)” in their sample of 54 chil- ative feelings upon reunion were openly redren of depressed and nondepressed mothers. solved. The behavior that contributed to the In the Canadian sample (Chisholm, 1996) Secure (other) classification and set her apart only 1 of the 42 children classified as secure from other secure children was her behavior in the comparison groups was classified as toward the stranger. The child initiated all inSecure (other), but 37.5% of the 16 secure or- teraction with the stranger, watched the stranger phanage children were classified as Secure quite openly, and at one point left the interac(other). This represents a much higher per- tion with the parent to search out the stranger. centage of Secure (other) children than is This is clearly not behavior that would be disfound in the literature on other high risk sam- played by a child classified as typically seples using the PAA. cure. However, we cannot assume that, beThis finding suggested the possibility that cause the behavior of orphanage children may the coding systems that were developed using manifest itself in somewhat different ways in normative samples of children were inade- the context of a separation–reunion procequate to evaluate attachment in institutional- dure, they are necessarily insecure. This is an ized samples, that is, that children classified important way in which orphanage children as Secure (other) might not truly be secure. have reshaped our thinking about the meaning Chisholm (1996) compared Secure (other) chil- of behavior in the context of procedures that dren to both the remaining secure children and were developed with normative samples. to children classified as Insecure (other) to Bowlby claimed that the development of a find out which of these groups they resembled first attachment relationship occurs sometime more. (It is important to note that all attach- during the second half of the first year of life ment coding was done by two independent (Bowlby, 1969/1982) and therefore it was only coders who were blind to children’s group after this time that separation from an attachmembership and all other behavior besides ment figure would be psychologically harmattachment.) Secure (other) children did not ful. He reported that infants who had been differ from secure orphanage children on be- adopted between 6 and 9 months of age often havior problems, parenting stress, or indiscrim- showed little or no socioemotional damage. inate friendliness, but they did have lower IQs The findings from orphanage samples concur than secure orphanage children. In contrast, with Bowlby’s theory and with studies from they differed greatly from children classified noninstitutionalized samples. In all three Roas Insecure (other), scoring lower on behavior manian studies, children who were removed Institutionalization and child development from the institution prior to the time when attachment is expected to develop (Chisholm’s (1998) “early-adopted” group, adopted before 4 months; Marcovitch et al.’s (1997) “home” group, adopted before 6 months; and O’Connor et al.’s (2003) “early-placed” group, adopted before 6 months) were no more likely to develop insecure attachment relationships than children who were never institutionalized. Institutionalization, even under the extremely poor conditions found in Romania, does not appear to have an impact on attachment when it ends before the appropriate time for such development. Thus, there is theoretical and empirical agreement concerning the onset of a possible sensitive period for attachment. Bowlby (1969/1982) felt that a child was maximally ready to become attached quickly at least through the end of the first year, but thereafter attachment would be more difficult (e.g., require a longer time or better parenting or take a less adaptive form). In both the Canadian and U.K. samples (Chisholm, 1998; O’Connor et al., 2003), children who had experienced more than 6 months of institutionalization, that is, children who remained in institution without a primary caregiver at the time when a first attachment would have been expected to develop, were more likely to be insecurely attached than comparison group children. In other areas of child development the impact of institutionalization has been shown to depend on how long a child has spent in the institution. In terms of attachment this association has not been found. Chisholm et al. (1995) did not find a relationship between orphanage children’s age at adoption and parents’ reports of their children’s attachment security 11 months postadoption. In a follow-up study using a separation reunion procedure to measure attachment patterns when children were 4.5 years or older, previously institutionalized children classified as insecure were no more likely than previously institutionalized children classified as secure to have spent a longer time in institution or to have been older when they were adopted (Chisholm, 1998). Thus, examination within the orphanage group did not reveal an association between children’s attachment patterns and the length of 875 time they had spent in institution. This agrees with Tizard and Hodges (1978), who did not find a relationship between whether a parent considered his or her child to be attached at 8 years of age and the length of time the child had spent in institution: the majority of children who had spent at least 4.5 years in institution were reported to be closely attached to their parent at 8 years of age. Given the lack of a general relationship between length of institutionalization and the quality of relationships that children later form with their adoptive parents, it remains unclear whether there is a length of time beyond which developing a secure attachment relationship may become impossible. Bowlby initially claimed that if the opportunity to form an attachment was delayed until after the age of 2.5 years, it was “almost useless” (Bowlby, 1953), thereby implying a short window of opportunity for developing attachment. As Bowlby’s theory developed, the length he attributed to this sensitive period increased. By 1973, he claimed that the sensitive period for attachment likely extended through the decade beyond a child’s fifth birthday, but that such development would become increasingly more difficult as a child grew older. This was in line with a more modern emphasis on risk and resilience in a probabilistic, rather than deterministic, model of development (Rutter & O’Connor, 1999). Ames and Chisholm (2001) have noted that the statistical analyses used in most studies do not enable an examination of such a “sensitive period” hypothesis. They claim that such a hypothesis is better examined by a careful inspection of data rather than by either correlations (that investigate how well data fit a straight line) or average differences between age groups. In a reanalysis of the Chisholm (1998) data, Ames and Chisholm (2001) found that the median and ranges of length of institutionalization were 16.5 (range = 9–39 months) for orphanage children subsequently classified as secure; 14 (range = 9–53) months for children classified as typical insecure; and 21 (range = 8–53) months for children classified as atypical insecure. These differences were not statistically significant, but it is important to note that the oldest children who had de- 876 K. MacLean veloped a secure attachment relationship with these children scored above the clinical cutoff an adoptive parent had spent only 39 months on behavior problems, and three of the four in an orphanage. In this sample there were for whom parenting stress scores were availonly 5 children who had spent more than 39 able had higher than average parenting stress months in an orphanage. All of those children scores for the orphanage group. Three of them had developed an insecure attachment rela- had a sibling who was adopted from Romania tionship, and 4 of the 5 had an atypical inse- at the same time. All of these factors would cure attachment. It is clear that on the basis of undoubtedly compromise their parents’ ability 5 children no firm conclusions can be drawn to respond sensitively to them. Therefore, concerning the end of a sensitive period for rather than assume that length of institutionalthe development of a first attachment. More- ization per se explains their lack of secure atover, the insecure attachments that children tachment, it appears more reasonable to sugdeveloped were also associated with particu- gest that they were adopted into families in lar factors of both the child and the parent, which the resources were not sufficient to suggesting that the insecurity of attachment allow parents to provide children who had might be associated with those factors rather many problems with the high level of sensithan with any upper limit on the timing of a tive responsiveness that would be required to period for the formation of a first attachment. develop a secure attachment. Characteristics of both the child and the The suggestion that stressors that interfere parent have been associated with attachment with parents’ sensitive responsiveness negasecurity. Children with lower IQs (Chisholm, tively affect children’s attachment fits with 1998; Tizard & Hodges, 1978) and more be- Tizard and Hodges’ (1978) findings with a havior problems (Chisholm, 1998; Marcovitch group of children who had been admitted to et al., 1997) have more difficulty forming at- institution before 4 months of age and had retachment relationships with their adoptive par- mained there until at least two years of age. ents, most likely because these factors in- Between 2 and 4 years of age, 24 of the chilterfere with parents’ ability to be sensitively dren were adopted out of institution and 15 responsive to their children. Ames and Chis- children were restored to their natural famiholm (2001) found another factor that may lies. In comparing the adoptive and restored have interfered with parents’ ability to be sen- groups of children, Tizard and Hodges (1978) sitively responsive. Although only 14% of se- found that whereas 84% of adoptive mothers curely attached and typical insecurely attached felt that their child was deeply attached to children were adopted by families that had them, only 54% of mothers of restored chiladopted another Romanian child at the same dren felt this was the case. Children in the time, 57% of children classified as Atypical restored group were reunited with their bioInsecure had a sibling who had been adopted logical mothers who often were ambivalent or from Romania at the same time. Additional reluctant to have their child return home. When family factors that were associated with chil- compared to children in the adopted group, dren’s insecure attachment patterns were lower restored children typically were returned to socioeconomic status and a higher level of families who had a larger number of children parenting stress (Chisholm, 1998). These stress- (Tizard & Hodges, 1978), whose mothers were ors are among those believed by Belsky (1999) younger and whose fathers had jobs of a to have a negative impact on parents’ ability lower socioeconomic status (Tizard & Rees, to be sensitively responsive to their infants. 1974). Parents of restored children also spent These factors may help to explain the inse- less time in play and educational activities cure attachments of the oldest five children in with their children than parents in the adopthe Ames and Chisholm (2001) reanalysis. All tive group (Hodges & Tizard, 1989). In confive of these children had IQ scores lower trast, adoptive parents wanted their children than 85, and four of the five came from fami- and devoted a great deal of time to them. Tizlies whose income was lower than the average ard and Hodges (1978) concluded that formfor the orphanage group as a whole. Three of ing a secure attachment relationship did not Institutionalization and child development so much depend on institutionalization as on the “willingness of the new parents to accept a dependent relationship and to put a lot of time and effort into developing it” (p. 115). To summarize: it is very rare to find children who have received so little one-on-one attention from adults as children reared in the extremely deprived environment of orphanages. Therefore, studies of institutionalized children have provided researchers with a unique opportunity to examine the development of a first attachment beyond the time it appears in family-reared children. Early researchers disagreed as to whether previously institutionalized children were able to form an attachment relationship with their adoptive parents, but this disagreement was likely the result of conditions in the institution and the post institutional environment, both of which varied widely across studies. Goldfarb (1943a, 1945b) provided the most pessimistic view by claiming that after the experience of early institutionalization children were incapable of forming attachment relationships. He, however, studied children who were housed in very poor-quality institutions and who subsequently experienced several foster home placements. In contrast, Tizard and Hodges (1978) were optimistic and claimed that even after 4.5 years in institution children were capable of forming attachment relationships. They, however, studied children from orphanages in which the conditions were far superior to the conditions found in Romania, and the children in Tizard’s studies were typically adopted into stable homes in which parents were highly invested in them. Recent studies of attachment in previously institutionalized Romanian children are consistent in indicating that after children have been in institution for many months without an opportunity to form an attachment relationship they are able to form attachments with their adoptive parents, and that furthermore, a third or more of these attachments are secure. Attachments in previously institutionalized children may be slower to develop, and a higher percentage of orphanage children than familyreared children form insecure attachment relationships, but almost two-thirds of children from orphanages are able to form attachments 877 that are similar to those found in normative samples. The remaining third show quite atypical attachment patterns, and the types of atypical patterns they display differ from other maltreated samples. It is worth noting the value of the Secure (other) classification that is included in Crittenden’s (1992b) PAA system. Although coders were blind to children’s group membership, they were able to discriminate between the Secure (other) and Insecure (other) patterns. This provides construct validity for the PAA system and accentuates the importance of the Secure (other) pattern in examining attachment in atypical samples. Even when children have atypical ways of showing a secure attachment, they can be accurately coded as secure. The length of institutionalization for children who remained in institution beyond the first half of the first year of life does not appear to be related to the quality of the attachment relationship that they subsequently formed. Chisholm (1998) found that after 3.5 years or more of orphanage life, children were able to form attachments with their adoptive parents. One child who spent 39 months in orphanage had developed a secure attachment, and another child who had spent 53 months in orphanage had developed a typical insecure attachment. The majority of adoptive parents in Tizard’s sample whose children had spent 4.5 years in orphanage claimed their child was deeply attached to them a few years after adoption (Tizard & Hodges, 1978). We do not yet know whether the period during which attachment is possible might be extended through adolescence, as Bowlby (1988) has suggested, or whether there is an earlier time after which the development of attachment becomes impossible. It is probably better at present to conceive of attachment formation as being related to factors that compromise the ability of parents to be sensitively responsive to their child, rather than to some age-based offset of a sensitive period for formation of a first attachment. Such factors include the child’s lower IQ, more behavior problems, and the adoption of more than one child at a time. High levels of parenting stress and lower socioeconomic status also compromise parents’ ability to 878 handle children’s problems. All of these influences on the attachment relationship can be considered using a transactional argument (Cicchetti, 1996; Sameroff, 1983). As a result of the orphanage experience, children arrived in their adoptive homes with medical, physical, intellectual, and socioemotional problems that undoubtedly led to more stress for their parents; such stress interfered with parents’ ability to respond sensitively to their child’s cues, which led to more problems on the part of children and the further compromise of the attachment relationship. It could be argued then that it takes more than “good enough” parenting to promote secure attachment in previously institutionalized children, particularly when they present with a myriad of problems and family resources are limited in terms of dealing with such problems. Discussion The methodological problems of studies of institutionalized children are many and complex. Measures that were standardized on nonclinical or even clinical populations sometimes are not appropriate for the institutionalized sample. In the study of Romanian orphans adopted to Canada it has sometimes proven necessary to go to the level of subscales or even individual items to clarify what is happening. This creates a problem of multiple comparisons, and it is certainly not recommended that researchers routinely look at group differences on all items. But when there is a reason for separating items that have been grouped together in measures standardized on noninstitutional populations, the researcher must go beyond merely reporting the scale scores yielded by the standardized test. For example, Mainemer et al. (1998) found that adoptive parents scored their previously institutionalized children higher on the Distractibility/ Hyperactivity subscale of the Parenting Stress Index (Abidin, 1990) than did parents of Canadian-born children. There was nothing in parent interviews or in home visitors’ observations, however, that pointed to orphanage children being hyperactive (indeed, they were generally described as being fairly passive K. MacLean and quiet); and when the index Distractibility/ Hyperactivity subscale was divided into two sets of items, it was found that parents of orphanage children perceived them as more distractible but not more hyperactive. Later, as the children spent more time in their adoptive homes, they added hyperactivity to the distractibility. This result helped to explain what appeared to be contradictory findings in the earlier literature, with previously institutionalized children having been reported to be both very passive and very active (Goldfarb, 1943b, 1945a). A similar example of the poor fit of a standardized measure was found when the CBCL was used to rate orphanage children. Fisher et al. (1997) found that 11 months after adoption orphanage children scored higher on Internalizing, but not on Externalizing, than did comparison children. At 3 years after adoption, however, they scored higher on Externalizing but not Internalizing (Ames, 1997). Because observers and parents had reported that the children still had several unusual behaviors, the individual items that made up the Internalizing scale at the early age were traced and it was found that at the later testing the orphanage children still showed their earlier specific “internalizing” behaviors (stares blankly, strange behavior, acts too young, speech problems) even though those items had shifted onto different subscales in the CBCL version used to test older children. Although standardized measures require care when applied to institutionalized children, studying variables that do not have standardized measures requires the development of new measures that lack proven validity. Such was the case with measures of Indiscriminate friendliness, which is a strong characteristic of orphanage children. Both the Canadian study (Chisholm, 1998; Chisholm et al., 1995) and the U.K. study (O’Connor et al., 1999, O’Connor, Rutter, et al., 2000) developed measures of descriptively similar behavior, but the measures are slightly different and have been given different names (Chisholm’s “indiscriminate friendliness” vs. O’Connor’s “disinhibited attachment disturbance”), which indicate the researchers’ preferences for the Institutionalization and child development measure’s theoretical relationship or lack of relationship to attachment. The difference in theoretical emphasis between the measures is likely to generate fruitful research in the future. Although there is the temptation to claim that samples of previously institutionalized children can address questions concerning sensitive periods in development, the answers that can be provided are limited. Ames and Chisholm (2001) have discussed the difficulties of trying to use studies of postinstitutionalized children to demonstrate the presence or absence of sensitive periods. To prove that a sensitive period exists it is necessary to show both the exact ages that are important and the length of time the operative factor (in this case, institutionalization) must be in place for effects to occur. To conclude that it is deprivation during a particular age period that matters, it is necessary to demonstrate not only that a shorter duration of deprivation is insufficient to produce the same effect but also that the same duration would not have the same effect at another time of life. This requires studies that systematically vary not only the age at which deprivation begins but also the duration of deprivation at each of those ages, something that does not occur in experiments in nature. In spite of the challenges of interpretation and the methodological differences among studies of previously institutionalized children, the results across studies are consistent in showing that institutionalization has a powerful impact on all aspects of children’s development. When compared to either other adopted children or children home-reared since birth, orphanage children have lower IQs, are shorter and weigh less, and have more behavior problems and attention difficulties. They are also more indiscriminately friendly and have more insecure attachments. There is not an area in which orphanage children remain unscathed. What is yet to be specified are the antecedent factors inherent in an institutional upbringing that explain these powerful outcomes. There is some evidence that the quality of the institution makes a difference to developmental outcomes. Morison and colleagues (1995) found 879 that the availability of toys and having been a favorite in the institution were associated with fewer delays among orphanage children, whereas having been described as dirty when first met by parents was associated with more delays. It is unclear, however, whether one or more institutional factors explain delays in all areas or whether there are specific factors that are associated with outcomes in specific areas. For example, intellectual delay may be explained either by the impact of malnutrition on brain development or by the lack of cognitive stimulation in the orphanage environment or by both factors operating together. Given the profound general deprivation that often characterizes orphanages, researchers have been unable to specify which aspects of deprivation caused the outcomes that have been found. The only exception is Tizard’s work (1977) with orphanage children in the United Kingdom, who experienced only social emotional deprivation but were well cared for otherwise. We know from this work that providing good nutrition and cognitive stimulation can prevent developmental delays. These same children, however, still displayed indiscriminate friendliness and some difficulties in attachment, which seem attributable to the limitation of not having had a close personal relationship with an adult caregiver. Presently we simply do not know enough about conditions in orphanages. Careful observational studies of children’s lives in different orphanage settings are sorely needed in order to identify more clearly the particular factors in the orphanage environment that have an impact on developmental outcomes. Future research studying samples of children from orphanages in other countries where conditions differ in nutrition, access to children of different ages, child to caregiver ratios, attitudes of caregivers toward children, and number of toys available, may be better able to isolate particular antecedents that explain specific post institutionalization outcomes. At the same time it is important for current longitudinal studies to continue so that the long-term effects of early institutionalization can be examined. Current studies are presently reporting findings 10 to 11 years after 880 K. MacLean institutionalization. This work will help us to dren still housed in institutions around the differentiate developmental delay from per- world. For example, future research must fomanent damage and thus extend our knowl- cus on examining any possible interventions edge concerning long term prognoses for that put human interaction back in the lives of postinstitutionalized children. orphanage children. Improving child to careA second focus for future research lies in giver ratios in orphanages is often a financial intervention studies. Research efforts should impossibility, but such improvement might be focus on identifying the kinds of interventions realized by either the careful use of volunteers that could be implemented in the orphanage from the community or by providing service context to ameliorate negative developmental learning experiences for students who are inoutcomes. There are presently two interven- terested in child development and education. tion studies ongoing in Eastern Europe. The This could provide a practical solution to the Bucharest Early Intervention Project (Koga, problem of high child to caregiver ratios in Smyke, Zeanah, 2003; Zeanah, Smyke, & Koga, institutions, but it would have to be imple2003) has begun providing a foster care inter- mented with care to ensure that such volunteers vention in Bucharest, Romania, in which they were committed and somewhat stable figures are following the development of children re- in children’s lives. Children in Eastern Euromoved from orphanage to foster care and com- pean orphanages are typically housed in age paring their developmental progress to both segregated groups, a policy that hinders cogchildren who remain in orphanage and a com- nitive growth. More than 50 years ago Skodak munity comparison group. The St. Petersburg– and Skeels (1945, 1949) showed that placing U.S.A. Orphanage Project (McCall, Muha- institutionalized infants as “house guests” with medrahimov, Groark, Palmov, & Nikiforova, older residents in the institution improved in2003), operating in three baby homes in St. fants’ cognitive competence, so the idea of Petersburg, is providing different interven- age integration is not new but is a change that tions to examine which are most effective in could be implemented fairly easily and cost reducing the developmental impact of institu- effectively. This is one aspect of the interventional rearing. In one baby home, orphanage tion being conducted by the St. Petersburg– staff are being trained in child development U.S.A. Orphanage Project. If young children and sensitive responsiveness. In a second, were housed with other children who were both caregiver training and structural changes slightly above their own developmental level, that reduce the number of different caregivers this would offer them a zone of proximal deand increase their stability in the lives of the velopment that could promote positive develchildren are being implemented. A third baby opmental outcomes (Vygotsky, 1978). home is serving as a “no intervention” comPiaget’s constructivist theory emphasizes parison group. Preliminary reports have shown that interacting with both objects and people that these interventions have been successful stimulates cognitive development. Both kinds in improving both caregivers’ attitudes and of interaction are sorely lacking in an orphanbehavior toward the children in their care and age environment. Future interventions promotchildren’s scores on standardized measures of ing cognitive development could provide orpersonal–social, communication, and cogni- phanage children with toys and/or computer tive skills. games that would allow children to make someThese projects hold great promise for spec- thing happen in the environment. This would ifying the kinds of interventions that may be provide them with a sense of personal effectmost effective at ameliorating the develop- ance and stimulate cognitive growth. Peer inmental delays so common in orphanage chil- teraction in orphanages might be increased dren. Given their high cost, however, future through the use of social interactive toys that research should also examine whether there would require two or more children working are less expensive interventions that might be together to operate them. If researchers could implemented on a wider scale as a partial so- demonstrate that toys that required such social lution for the hundreds of thousands of chil- interaction actually improved developmental Institutionalization and child development outcomes then they could be implemented on a wider scale than more expensive interventions. Research on institutionalized children has shown that similar experiences early in life may result in a variety of outcomes, a phenomenon that Cicchetti (1996) has termed multifinality. We know that orphanage life and the intellectual, physical, social, and emotional deprivation that it entails is clearly a risk factor for less than optimal development. What we have learned from studies of institutionalized children, however, is that having experienced institutionalization does not necessarily doom a child to developmental insult. It also matters what happens after leaving institution. On some measures poorer outcomes are associated not only with early institutional rearing but also 881 with family and parent characteristics in the adoptive home. A stimulating and supportive home environment was associated with higher IQs in orphanage children (Morison et al., 1995), and children’s insecure attachment patterns were associated with parents having lower socioeconomic status and more stress (Chisholm, 1998). Institutionalization is clearly a risk factor for compromised development, but it is not possible to predict developmental outcome with any certainty knowing only that a particular child has been institutionalized early in life. When institutionalization is combined with other risk factors (e.g., low IQ, behavior problems, parenting stress, low socioeconomic status), it becomes easier to predict poor developmental outcomes. References Abidin, R. R. (1990). Parenting Stress Index (3rd Ed.). Charlottesville, VA: Pediatric Psychology Press. Achenbach, T. W., Edelbrock, C., & Howell, C. T. (1987). Empirically based assessment of the behavioral/emotional problems of 2–3-year-old children. Journal of Abnormal Child Psychology, 15, 629–650. Ahadi, S. A., Rothbart, M. K., & Ye, R. (1993). Child temperament in the US and China: Similarities and differences. European Journal of Psychology, 7, 359– 377. Ainsworth, M. D. A., Blehar, M. C., Waters, E., & Wall, S. (1978). Patterns of attachment: A psychological study of the Strange Situation. Hillsdale, NJ: Erlbaum. Ames, E. W. (1997). Development of Romanian orphanage children adopted to Canada. Final Report to the Human Resources Development Office, Ottawa, Canada. Ames, E. W., & Chisholm, K. (2001). Social and emotional development in children adopted from institutions. In D. B. Bailey, Jr., J. T. Bruer, F. J. Symons, & J. W. Lichtman (Eds.), Critical thinking about critical periods (pp. 129–148). Baltimore, MD: Brookes. Bayley, N. (1969). The Bayley Scales of Infant Development: Birth to two years. San Antonio, TX: Psychological Corporation. Beckett, C., Bredenkamp, D., Castle, J., Groothues, C., O’Connor, T.G., Rutter, M., & the English and Romanian Adoptees Study Team. (2002). Behavior problems associated with institutional deprivation: A study of children adopted from Romania. Journal of Developmental and Behavioral Pediatrics, 23, 297–303. Belsky, J. (1999). Interactional and contextual determinants of security. In J. Cassidy & P. R. Shaver (Eds.), Handbook of attachment: Theory, research and clinical applications (pp. 249–264). New York: Guilford Press. Benoit, T. C., Jocelyn, L. J., Moddeman, D. M., & Embree, J. E. (1996). Romanian adoption: The Manitoba Experience. Archives of Pediatric and Adolescent Medicine, 150, 1278–1282. Bowlby, J. (1951). Maternal care and mental health (WHO Monograph No. 2). Geneva: World Health Organization. Bowlby, J. (1953). Child care and the growth of love. Baltimore, MD: Penguin Books. Bowlby, J. (1982). Attachment and loss: Vol. 1. Attachment. New York: Basic Books. (Original work published 1969) Bowlby, J. (1988). Lecture 2: The origins of attachment theory. A secure base (pp. 20–38). New York: Basic Books. Broussard, M., & DeCarie, T. G. (1971). The effects of three kinds of perceptual-social stimulation on the development of institutionalized children: Preliminary report of a longitudinal study. Early Child Development and Care, 1, 111–130. Caldwell, B. M., & Bradley, R. H. (1984). Home observation for measurement of the environment. Unpublished manuscript, University of Arkansas at Little Rock. Carlson, E. A., & Sroufe, A. L. (1995). Contributions of attachment theory to developmental psychopathology. In D. Cicchetti & D. J. Cohen (Eds.), Developmental psychopathology: Vol. 1. Theory and methods (pp. 581–617). New York: Wiley. Carlson, M., & Earls, F. (1997). Psychological and neuroendocrinological sequelae of early social deprivation in institutionalized children in Romania. Annals of the New York Academy of Sciences, 807, 419–428. Carlson, V., Cicchetti, D., Barnett, D., & Braunwald, K. (1989). Finding order in disorganization: Lessons from research on maltreated infants’ attachments to their caregivers. In D. Cicchetti & V. Carlson (Eds.), Child maltreatment: Theory and research on the causes and consequences of child abuse and neglect (pp. 494–528). New York: Cambridge University Press. Cassidy, J., Marvin, R. S., & the MacArthur Working 882 Group (1992). Attachment organization in preschool children: Procedures and coding manual. Unpublished manuscript, University of Virginia. Chisholm, K. (1996). Attachment security and indiscriminately friendly behavior in children adopted from Romanian orphanages. Doctoral dissertation. Simon Fraser University, Burnaby, British Columbia, Canada. Chisholm, K. (1998). A three-year follow-up of attachment and indiscriminate friendliness in children adopted from Romanian orphanages. Child Development, 69, 1092–1106. Chisholm, K. (2000). Attachment in children adopted from Romanian orphanages: Two case studies. In P. M. Crittenden & A. H. Claussen (Eds.), The organization of attachment relationships: Maturation, culture, and context (pp. 171–189). New York: Cambridge University Press. Chisholm, K., Carter, M., Ames, E. W., & Morison, S. J. (1995). Attachment security and indiscriminately friendly behavior in children adopted from Romanian orphanages. Development and Psychopathology, 7, 283–294. Chugani, H. T., Behen, M. E., Muzik, O., Juhász, C., Nagy, F., & Chugani, D. C. (2001). Local brain functional activity following early deprivation: A study of postinstitutionalized Romanian orphans. Neuroimage, 14, 1290–1301. Cicchetti, D. (1996). Child maltreatment: Implications for developmental theory and research. Human Development, 39, 18–39. Cicchetti, D., & Barnett, D. (1991). Attachment organization in maltreated preschoolers. Development and Psychopathology, 3, 397–411. Crittenden, P. M. (1985). Maltreated infants: Vulnerability and resilience. Journal of Child Psychology and Psychiatry, 26, 85–96. Crittenden, P. M. (1988). Relationships at risk. In J. Belsky & T. Nezworski (Eds.), Clinical implications of attachment (pp. 137–174). Hillsdale, NJ: Erlbaum. Crittenden, P. M. (1992a). Quality of attachment in the preschool years. Development and Psychopathology, 4, 209–243. Crittenden, P. M. (1992b). Children’s strategies for coping with adverse home environments: An interpretation using attachment theory. Child Abuse and Neglect, 16, 329–343. Crittenden, P. M., & Claussen, A. H. (2000). Adaptation to varied environments. In P. M. Crittenden & A. H. Claussen (Eds.), The organization of attachment relationships: Maturation, culture, and context (pp. 234– 248). New York: Cambridge University Press. Crittenden, P. M., Partridge, M. F., & Claussen, A. H. (1991). Family patterns of relationship in normative and dysfunctional families. Development and Psychopathology, 3, 491–512. Dennis, W. (1973). Children of the crèche. New York: Appleton–Century–Croft. Dettling, A., Feldon, J., & Pryce, C. R. (2002). Repeated parental deprivation in the infant common marmoset (Callithrix jacchus, Primates) and analysis of its effects on early development. Biological Psychiatry, 52, 1037–1046. DeWolff, M. S., & van IJzendoorn, M. H. (1997). Sensitivity and attachment: A meta-analysis on parental antecedents of infant attachment. Child Development, 68, 571–591. Fernyhough, L., Audet, K., & Le Mare, L. (2002, August 2–5). Attachment in Romanian orphans ten years K. MacLean after adoption to Canada. Poster presented at the International Society of Social and Behavioral Development, Ottawa, Canada. Fisher, L., Ames, E. W., Chisholm, K., & Savoie, L. (1997). Problems reported by parents of Romanian orphans adopted to British Columbia. International Journal of Behavioral Development, 20, 67–83. Flint, B. M. (1978). New hope for deprived children. Toronto: University of Toronto Press. Frankenberg, W. K. (1986, May). Revised Denver Prescreening Developmental Questionnaire. Paper presented in part at the Ambulatory Pediatric Association meeting, Washington, DC. Goldfarb, W. (1943a). The effects of early institutional care on adolescent personality. Journal of Experimental Education, 12, 106–129. Goldfarb, W. (1943b). Infant rearing and problem behavior. American Journal of Orthopsychiatry, 13, 249– 265. Goldfarb, W. (1944). Infant rearing as a factor in foster home replacement. American Journal of Orthopsychiatry, 14, 162–166. Goldfarb, W. (1945a). Psychological privation in infancy and subsequent adjustment. American Journal of Orthopsychiatry, 14, 247–255. Goldfarb, W. (1945b). Effects of psychological deprivation in infancy and subsequent stimulation. American Journal of Psychiatry, 102, 18–33. Goldfarb, W. (1947). Variations in adolescent adjustment of institutionally-reared children. American Journal of Orthopsychiatry, 17, 449–457. Goldfarb, W. (1955). Emotional and intellectual consequences of psychologic deprivation in infancy: A reevaluation. In P. Hoch & J. Zubin (Eds.), Psychopathology of Childhood (pp. 105–119). New York: Grune & Stratton. Greenberg, M. T. (1999). Attachment and psychopathology in childhood. In J. Cassidy & P. R. Shaver (Eds.), Handbook of attachment: Theory, research, and clinical applications (pp. 469–496). New York: Guilford Press. Groza, V. & Ileana, D. (1996). A follow-up study of adopted children from Romania. Child and Adolescent Social Work Journal, 13, 541–565. Gunnar, M. R. (2000). Early adversity and the development of stress reactivity and regulation. In C. A. Nelson (Ed.), Minnesota Symposium on Child Psychology: Vol. 31. The effects of early adversity on neurobehavioral development (pp, 163–200). Mahwah, NJ: Erlbaum. Gunnar, M. R., Morison, S. J., Chisholm, K., & Shuder, M. (2001). Salivary cortisol levels in children adopted from Romanian orphanages. Development and Psychopathology, 13, 611–628. Gyorkos, T. W., & MacLean, J. D. (1992). Medical evaluation of children adopted from abroad. New England Journal of Medicine, 268, 410. Heim, C., Owen, M. J., Plotsky, P. M., & Nemeroff, C. B. (1997). The role of early adverse life events in the etiology of depression and posttraumatic stress disorder: Focus on corticotrophin-releasing factor. Annals of the New York Academy of Science, 821, 194–207. Hodges, J., & Tizard, B. (1989). Social and family relationships of ex-institutional adolescents. Journal of Child Psychology and Psychiatry, 30, 77–97. Hogg, C., Rutter, M., & Richman, N. (1997). Emotional and behavioral problems in children. In I. Sclare Institutionalization and child development (Ed.), Child psychology portfolio. Windsor, Canada: NFER–Nelson. Hoksbergen, R.A.C. (1981). Adoption of foreign children to the Netherlands. International Child Welfare Review, 49, 28–37. Hostetter, M. K., Iverson, S., Thomas, W., McKenzie, D., Dole, K., & Johnson, D. E. (1991). Medical evaluation of internationally adopted children. New England Journal of Medicine, 325, 479–485. Hunt, J. M., Mohandessi, K., Ghodessi, M., & Akiyama, M. (1976). The psychological development of orphanage-reared infants: Interventions with outcomes (Tehran). Genetic Psychological Monographs, 94, 177–226. Jenista, J. A. (1992). Disease in adopted children from Romania [Letter to the editor]. Journal of the American Medical Association, 268, 601. Johnson, D. E., Miller, L. C., Iverson, S., Thomas, W., Franchino, B., Dole, K., Kiernan, M. T., Georgieff, M. K., & Hostetter, M. K. (1992). The health of children adopted from Romania. Journal of the American Medical Association, 268, 3446–3451. Kaler, S. R., & Freeman, B. J. (1994). Analysis of environmental deprivation: Cognitive and social development in Romanian orphans. Journal of Child Psychology and Psychiatry, 35, 769–781. Knobloch, H., Stevens, F., & Malone, A. F. (1980). Manual for developmental diagnosis: The administration and interpretation of the revised Gesell and Amatruda developmental and neurologic examination. New York: Harper & Row. Koga, S., Smyke, A. T., & Zeanah, C. H. (2003, April), Effects of foster care on physical and language development following institutionalization. Paper presented at the Biennial Meeting of the Society for Research in Child Development, Tampa, FL. Kreppner, J. M., O’Connor, T. G., & Rutter, M. (2001). Can inattention/overactivity be an institutional deprivation syndrome? Journal of Abnormal Child Psychology, 29, 513–528. Le Mare, L., & Audet, K. (2002, August 2–5). Attention abilities of Romanian orphans ten years after being adopted to Canada. Poster presented at the International Society for the Study of Behavioral Development, Ottawa, Ontario. Le Mare, L., Vaughan, K., Warford, L., & Fernyhough, L. (2001, April 19–22). Intellectual and academic performance of Romanian orphans 10 years after being adopted to Canada. Poster presented at the Biennial Meeting of the Society for Research in Child Development, Minneapolis, MN. Lieberman, A. F., & Paul, J. H. (1988). Clinical applications of attachment theory. In J. Belsky & T. Nezworski (Eds.), Clinical implications of attachment (pp. 327–347). Hillsdale, NJ: Erlbaum. Lieberman, A. F., & Zeanah, C. H. (1995). Disorders of attachment in infancy. Infant Psychiatry, 4, 571–587. Longstreth, L. E. (1981). Revisiting Skeels’ final study: A critique. Developmental Psychology, 17, 620–625. Lowrey, L. G. (1940). Personality distortion and early institutional care. American Journal of Orthopsychiatry, 10, 576–585. Mainemer, H., Gilman, L. C., & Ames, E. W. (1998). Parenting stress in families adopting children from Romanian orphanages. Journal of Family Issues, 19, 164–180. Marcovitch, S., Cesaroni, L., Roberts, W., & Swanson, C. 883 (1995). Romanian adoption: Parents’ dreams, nightmares, and realities. Child Welfare, 74, 993–1017. Marcovitch, S., Goldberg, S., Gold, A., Washington, J., Wasson, C., Krekewich, K., & Handley–Derry, M. (1997). Determinants of behavioral problems in Romanian children adopted to Ontario. International Journal of Behavioral Development 20, 17–31. McCall, R. B., Muhamedrahimov, R. J., Groark, C. J., Palmov, O. I., & Nikiforova, N. V. (2003, April). Research design and measurements in the St. Petersburg–U.S.A. orphanage project. Paper presented at the Biennial Meeting of the Society for Research in Child Development, Tampa, FL. McCarthy, D. (1972). The McCarthy Scales of Children’s Abilities. New York: Psychological Corporation/Harcourt Brace Jovanovich. McEwen, B. (1998). Stress, adaptation, and disease: Allostasis and allostatic load. Annals of the New York Academy of Science, 840, 33–44. Morison, S. J., Ames, E. W., & Chisholm, K. (1995). The development of children adopted from Romanian orphanages. Merrill–Palmer Quarterly, 41, 411–430. Morison, S. J., & Ellwood, A. L. (2000). Resiliency in the aftermath of deprivation: A second look at the development of Romanian orphanage children. Merrill– Palmer Quarterly, 46, 717–737. O’Connor, T. G., Bredenkamp, D., Rutter, M., & the ERA Study Team. (1999). Attachment disturbances and disorders in children exposed to early severe deprivation. Infant Mental Health Journal, 20, 10–29. O’Connor, T. G., Marvin, R. S., Rutter, M., Olrick, J., Britner, P. A., & the ERA Study Team. (2003). Child–parent attachment following early institutional deprivation. Development and Psychopathology, 15, 19–38. O’Connor, T. G., Rutter, M., & the ERA Study Team. (2000). Attachment disorder behavior following early severe deprivation: Extension and longitudinal follow-up. Journal of the American Academy of Child and Adolescent Psychiatry, 39, 703–712. O’Connor, T. G., Rutter, M., Beckett, C., Keaveney, L., Kreppner, J. M., & the English and Romanian Adoptees Study Team. (2000). The effects of global severe privation on cognitive competence: Extension and longitudinal follow-up. Child Development, 71, 376– 390. Pinneau, S. (1955). The infantile disorders of hospitalism and anaclitic depression. Psychological Bulletin, 52, 429–451. Provence, S., & Lipton, R. C. (1962). Infants in institutions. New York: International Universities Press. Ripley, M. (1992, August). Summary of comments made by adoptive parents of Bulgarian orphanage children. Paper presented at the 17th North American Conference on Adoptable Children, Ottawa, Canada. Rutter, M., & the English and Romanian Adoptees Study Team. (1998). Developmental catch-up, and delay, following adoption after severe global early privation. Journal of Child Psychology and Psychiatry, 39, 465–476. Rutter, M., & O’Connor, T. G. (1999). Implications of attachment theory for child care policies. In J. Cassidy & P. R. Shaver (Eds.), Handbook of attachment: Theory, research, and clinical applications (pp. 823– 844). New York: Guilford Press. Sameroff, A. (1983). Developmental systems: Contexts and evolution. In P. H. Mussen (Ed.), Handbook of 884 child psychology, (4th Ed., pp. 237–294). New York: Wiley. Sanchez, M. M., Ladd, C. O., & Plotsky, P. M. (2001). Early adverse experience as a developmental risk factor for later psychopathology: Evidence from rodent and primate models. Development and Psychopathology, 13, 419–450. Skodak, M., & Skeels, H. M. (1945). A follow-up study of children in adoptive homes. Journal of Genetic Psychology, 66, 21–58. Skodak, M., & Skeels, H. M. (1949). A final follow-up study of one hundred adopted children. Journal of Genetic Psychology, 75, 85–125. Sloutsky, V. M. (1997). Institutional care and developmental outcomes of 6-and7-year old children: A contextual perspective. International Journal of Behavioral Development, 20, 131–151. Solomon, J., & George, C. (1999). The measurement of attachment security in infancy and childhood. In J. Cassidy & P. R. Shaver (Eds.), Handbook of attachment: Theory, research, and clinical applications (pp. 287–316). New York: Guilford Press. Spitz, R. (1945a). Hospitalism: An inquiry into the genesis of psychiatric conditions in early childhood. Psychoanalytic Study of the Child, 1, 53–74. Spitz, R. (1945b). Hospitalism: A follow-up report. Psychoanalytic Study of the Child, 2, 113–118. Teti, D. M., Gelfand, D. M., Messinger, D. S., & Isabella, R. (1995). Maternal depression and the quality of early attachment: An examination of infants, preschoolers, and their mothers. Developmental Psychology, 31, 364–376. Thelan, E. (1979). Rhythmical stereotypies in normal human infants. Animal Behavior, 27, 669–715. Thelan, E. (1981). Rhythmical behavior in infancy: An ethological perspective. Developmental Psychology, 17, 237–257. Thorndike, R. L., Hagen, E. P., & Sattler, J. M. (1986). Manual for the Stanford–Binet Intelligence Scale (4th ed.). Chicago: Riverside. Tizard, B. (1977). Adoption: A second chance. London: Open Books. Tizard, B., Cooperman, O., Joseph, A., & Tizard, J. (1972). Environmental effects on language development: A study of young children in long-stay residential nurseries. Child Development, 43, 337–358. Tizard, B., & Hodges, J. (1978). The effect of early institutional rearing on the development of eight-year-old children. Journal of Child Psychology and Psychiatry, 19, 99–118. Tizard, B., & Joseph, A. (1970). Cognitive development of young children in residential care: A study of children aged 24 months. Journal of Child Psychology and Psychiatry, 11, 177–186. Tizard, B., & Rees, J. (1974). A comparison of the effects of adoption, restoration, to the natural mother, and K. MacLean continued institutionalization on the cognitive development of four-year-old children. Child Development, 43, 92–99. Tizard, B., & Rees, J. (1975). The effect of early institutional rearing on the behavior problems and affectional relationships of four-year-old children. Journal of Child Psychology and Psychiatry, 16, 61–73. Tizard, J., & Tizard, B. (1971). The social development of two-year-old children in residential nurseries. In H. R. Schaffer (Ed.), The origins of human social relations (pp. 147–163). New York: Academic Press. van IJzendoorn, M. H., Dijkstra, J., & Bus, A. G. (1995). Attachment, intelligence, and language: A meta-analysis. Social Development, 4, 115–128. Verhulst, F., Althaus, M., & Versluis–Den Bieman, H. J. (1990a). Problem behavior in international adoptees: I. An epidemiological study. Journal of the American Academy of Child and Adolescent Psychology, 29, 94–103. Verhulst, F., Althaus, M., & Versluis–Den Bieman, H. J. (1990b). Problem behavior in international adoptees: II. Age at placement. Journal of the American Academy of Child and Adolescent Psychology, 29, 104– 111. Vorria, P., Rutter, M., Pickles, A., Wolkind, S., & Hobsbaum, A. (1998). A comparative study of Greek children in long-term residential care and in two-parent families: I. Social, emotional, and behavioral differences. Journal of Child Psychology and Psychiatry, 39, 225–236. Vygotsky, L. S. (1978). Mind in society. Cambridge, MA: Harvard University Press. Warford, L. (2002). Peer relationships of Romanian orphans 10 years after adoption to Canada: A longitudinal study. M.A. thesis, Simon Fraser University, Burnaby, Canada. Wasserman, A. L. (1987). Common behavioral disorders of childhood. In M. I. Gottlieb & J. E. Williams (Eds.), Textbook of developmental pediatrics (pp. 235–253). New York: Plenum Press. Wormelli, C. T., & Carter, D. E. (1990). Manual for the Canada Quick Individual Educational Test. White Rock, Canada: Edumetrics Ltd. Zeanah, C. H. (1996). Beyond insecurity: A reconceptualization of attachment disorders in infancy. Journal of Consulting and Clinical Psychology, 64, 42–52. Zeanah, C. H. (2000). Disturbances of attachment in young children adopted from institutions. Journal of Developmental and Behavioral Pediatrics, 21, 230– 236. Zeanah, C. H., Smyke, A. T., & Koga, S. (2003, April). Effects of foster care following institutionalization on cognitive development and behavior problems. Paper presented at the Biennial Meeting of the Society for Research in Child Development, Tampa, FL.