Abstract
One of the major questions of human development is how early experience impacts the course of development years later. Children adopted from institutional care experience varying levels of deprivation in their early life followed by qualitatively better care in an adoptive home, providing a unique opportunity to study the lasting effects of early deprivation and its timing. The effects of age at adoption from institutional care are discussed for multiple domains of social and behavioral development within the context of several prominent developmental hypotheses about the effects of early deprivation (cumulative effects, experience-expectant developmental programming, and experience-adaptive developmental programming). Age at adoption effects are detected in a majority of studies, particularly when children experienced global deprivation and were assessed in adolescence. For most outcomes, institutionalization beyond a certain age is associated with a step-like increase in risk for lasting social and behavioral problems, with the step occurring at an earlier age for children who experienced more severe levels of deprivation. Findings are discussed in terms of their concordance and discordance with our current hypotheses, and speculative explanations for the findings are offered.
Keywords: Age at adoption, institutional deprivation, orphanage, early experience, developmental programming
The first months and years of life have long been regarded as crucial to later development; infancy is characterized by rapid neurological and behavioral development, and while some children are remarkably resilient, deprivation at such early ages can be particularly damaging. Examining the lasting effects of atypical early experiences informs our understanding of what experiences are necessary for typical development to proceed, how early experience produces lasting effects, and whether later positive experiences can be remediating. Further, if developmental outcomes systematically vary based on when an experience occurs, it implies that not only are certain experiences necessary to development, but their effects are uniquely potent during certain critical time windows. These questions are addressed via both animal and human studies; this review utilizes studies of children who experienced institutional rearing to inform our understanding of the lasting effects of early experience, particularly with regard to several prominent theories of early development.
Animal models are often used to better understand the lasting effects of atypical early relationship experience. Studies of monkeys demonstrated that the drive for maternal comfort is strong, and being reared in isolation produces atypical behaviors like self-mutilation and stereotypies (i.e., repetitive or ritualistic movements, postures, or utterances). While isolation lasting 6 months produced extensive deficits, the effects were more marked after 12 months of isolation, and only partial recovery was possible after rehabilitation (Anderson & Mason, 1974; Harlow, 1958; Kraemer, 1992; Rowland, 1964, as cited in Novak & Harlow, 1975). Neuroanatomical and physiological changes are evident long after isolation ceases (Parker & Maestripieri, 2011). Furthermore, if typical experiences are provided before a period of isolation, effects on monkeys’ behavior are less extreme, pointing to the importance of experience in the first months or years of life (Harlow & Novak, 1973; Suomi, 2003).
Rodent research has advanced our understanding of potential mechanisms to account for the importance of early experience. During the first two weeks of life, rodents’ glucocorticoid responses (which are maladaptive at higher levels, and particularly harmful during this age span) are maintained at a low level through maternal licking and grooming (Gunnar, 2003; Gunnar, Fisher, & The Early Experience, Stress, and Prevention Network, 2006). Maternal licking and grooming have also been associated with epigenetic effects that are related to gene expression and neural function into adulthood (Meaney, 2010). Thus, sensitive maternal behavior might serve a protective role, particularly during an early sensitive period; in humans, the age of 6 to 12 months might constitute a similar period, but evidence is limited (Gunnar, 2003).
Animal models allow us to study early deprivation through experimental paradigms that eliminate many of the confounds in naturalistic studies, but there are limitations to this approach. Primate research has typically focused on extreme levels of deprivation that are rarely found outside scientific laboratories; these findings are less informative about levels of deprivation that would be found for humans, and often cannot distinguish between the effects of the absence of a primary caregiver or the complete lack of social experience (Parker & Maestripieri, 2011).
In human research, orphanage institutions constitute an unfortunate, but unique, early rearing environment that informs our understanding of early development and the influence of timing of early experience. While studies of institutionalization lack the experimental control available in animal studies, they represent a natural experiment of how development may proceed when the typical parent-child relationship is absent, but varying degrees of other social experience are available. Children leave institutions at different ages for more supportive environments (typically, adoptive homes), so institutionalization offers a prime opportunity to study the effects of timing of early experience; there are few other natural circumstances in which children experience such a drastic shift in their quality of care. Studying post-institutionalized (PI) children can shed light on whether the effects of institutionalization (and more generally, early deprivation) are hard-wired in the first years of life, or are subject to change with later experience.
When institutionalization was first studied, psychoanalytically oriented researchers like Bowlby pointed to the role of a lack of mothering (e.g., tender and affectionate care), particularly between 6 to 12 months of age, in producing long-term deficits in PI children (Bowlby, 1951). Bronfenbrenner (1979) noted that researchers who were trained in laboratory research, on the other hand, attributed difficulties to a lack of environmental stimulation, and more recently, both factors are thought to contribute to the lasting effects of institutionalization (van IJzendoorn et al., 2011). Bronfenbrenner concluded, based on (somewhat limited) empirical evidence, that children are most vulnerable to long term negative effects of institutionalization in the first 6 months of life, and this risk decreases with older ages of admission into institutions. Moving to a more developmentally supportive environment was thought to allow some of the negative effects of institutionalization to be avoided.
Post-Institutionalized Children
While the early studies that informed Bowlby and Bronfenbrenner’s ideas (e.g., Goldfarb, 1943, 1944, 1945, 1947, 1955; Spitz, 1945a, 1945b) called attention to the potentially damaging and lasting effects of residing in a depriving environment in the early years of life, many of these studies were methodologically weak; the institutional environments were not often described in much (or any) detail, and it was sometimes unclear how many children were tested, what assessments were used and at what ages (MacLean, 2003). More recently, the research base has expanded, and while some methodological problems persist, the field has advanced to a point where a number of conclusions can be stated with some certainty.
First, internationally adopted children1 typically have more behavior problems than non-adopted parent-reared children, although the effect sizes are small (Juffer & van IJzendoorn, 2005). Children adopted after 12 months of age are less securely attached to their parents than non-adopted children (d = 0.80; van den Dries, Juffer, van IJzendoorn, & Bakermans-Kranenburg, 2009). At the time of adoption, they lag substantially behind their non-adopted peers in height (d = −2.43), weight (d = −2.60), and head circumference (d = −2.39), but these differences decline after several years in the adoptive home (ds = −1.56 to −0.57; van IJzendoorn, Bakermans-Kranenburg, & Juffer, 2007). While gender differences are rarely assessed in this population, a meta-analysis detected no significant differences between male and female international adoptees for behavior problems (Juffer & van IJzendoorn, 2005).
Second, while the rates of problems are higher in PI children than in parent-reared children, most PI children fall in the normal range of adjustment; PI children often show remarkable resiliency, despite the challenges they faced early in life. When PI children have problems, they are most often in specific domains that are likely related to their early experience. PI children are more likely to demonstrate disinhibited social behavior; quasi-autism, or autistic-like features; inattention/overactivity; behavior problems, especially externalizing problems, stereotyped behavior, and eating problems; and cognitive impairment (Beckett et al., 2002; MacLean, 2003; Merz & McCall, 2010, 2011; Rutter et al., 2010; Sonuga-Barke et al., 2008; Verhulst, Althaus, & Versluis-Den Bieman, 1990; Vorria, Rutter, Pickles, Wolkind, & Hobsbaum, 1998a). Cognitive impairment may be a consequence of residing in a depriving environment, rather than a consequence of group rearing (Castle et al., 1999; Rutter, 2005b).
Third, the severity of deprivation relates to the extent to which early institutionalization influences later development. There is a clear trend in the literature for children adopted from qualitatively better Chinese and Korean orphanages to have more moderate effects including behavior problems, social problems, and inattention/overactivity (e.g., Dalen, 2001; Tan, Marfo, & Dedrick, 2007, 2010), and children from globally depriving 1990s Romanian institutions to have more significant and lasting effects including stunted physical growth, autistic-like features, and stereotyped behaviors along with behavior problems, social problems, and inattention/overactivity (e.g., Rutter et al., 2010). When institutions are improved (The St. Petersburg-USA Orphanage Research Team, 2008) or when children are moved from institutions to better quality foster care (Smyke, Zeanah, Fox, & Nelson, & Guthrie, 2010), developmental and behavioral outcomes improve as well.
Effects of the timing of early institutionalization have been a focus in the literature (e.g., Bowlby, 1951; Bronfenbrenner, 1979; Gunnar, 2001; MacLean, 2003); age at adoption is typically the most ready index of how long children were exposed to depriving circumstances and the ages at which they were exposed to deprivation. While some studies find that later-adopted children have a higher rate of problems than earlier-adopted children (e.g., Bruce, Tarullo, & Gunnar, 2009; Camras, Perlman, Wismer Fries, & Pollak, 2006; Kreppner et al., 2007; Rutter et al., 2010), other studies find no effect of a child’s age at adoption (e.g., Miller, Chan, Tirella, & Perrin, 2009; Rojewski, Shapiro, & Shapiro, 2000; Tizard & Hodges, 1978). The inconsistency in effects across studies may be due to factors including the severity of deprivation, the specific outcome domains and assessments, or the age at assessment.
Hypotheses of How Early Deprivation Affects Development
It is clear that early institutionalization affects children’s development and that some of these effects are evident when children are adopted into families, but it is less clear how these effects come to be. Research on PI children has often lacked a basis in developmental theory (McCall, 2011), but some researchers (e.g., Rutter, O’Connor, & the ERA Study Team, 2004) have proposed several potential explanations for the effects of institutionalization and its timing.
Cumulative Effects
The cumulative effects hypothesis (O’Connor, 2003), also referred to as the continuation of adversity hypothesis (Rutter et al., 2004) or the cumulative deficit hypothesis (Jensen, 1974), posits that early experiences affect later outcomes largely through the continuation of depriving circumstances. A change in the quality of the environment should be accompanied by a change in the developmental trajectory, with near complete recovery possible over time. The timing of experience should have minimal influence relative to the duration and intensity of the experience. The cumulative effects hypothesis was favored in this field in its earlier years (Rutter, 1981), and some early experience researchers (e.g., Clarke & Clarke, 2000; Schaffer, 2000) still favor this hypothesis.
Thus, duration of institutionalization should relate to a child’s outcomes, but the specific ages of exposure should not; accordingly, age at adoption is expected to relate linearly to outcomes. When children are adopted into supportive homes, near complete recovery should be possible; the most problems might be found soon after adoption, with problems declining over time as children adapt more to their improved environment. Variations in care quality in the institution and in the adoptive home should contribute to individual differences.
Developmental Programming
Developmental programming suggests that effects of experience are “programmed” into the brain during certain sensitive developmental periods. Experience during sensitive periods (when somatic structures are being built) is crucial, and later experience is likely to have only a limited effect on outcomes (Rutter et al., 2004). There are two distinct developmental programming hypotheses.
Experience-expectant developmental programming
The experience-expectant hypothesis (Greenough, Black, & Wallace, 1987; Rutter et al. 2004; also discussed within the sensitive periods hypothesis in O’Connor, 2003) proposes that certain experiences are required within specific sensitive periods of development in order for normal development to occur. No child who lacks this necessary experience will have typical later development, and the effects are “hard wired” so that later experience can have only a minimal effect. The experiences that are thought to be necessary for typical development occur in a broad range of expectable environments; variations within the normal range aren’t thought to have a meaningful effect.
It follows that this hypothesis might predict that the specific ages of exposure to institutionalization (and thus, the age at adoption) should be crucial to predicting outcomes; if children are institutionalized during a sensitive period and do not have the necessary expected experiences, permanent effects should result for all children. A step function might be predicted to emerge for the age at adoption effect. In other words, there is expected to be a “threshold” such that age at adoption after that point is associated with increased risk for problems with no further increase in risk with even later ages at adoption. More time in a supportive home environment, or an older age at assessment, should not be associated with fewer problems. The criteria for “necessary experience” are not specified by this hypothesis, but children reared in institutions might lack an early caregiver-child relationship, care that is contingent on the child’s behavior, opportunity to freely explore the environment, or, in some cases, proper nutrition and medical care. It is possible that there will be a threshold of severity of deprivation such that only children who are “sufficiently deprived” would demonstrate effects.
Experience-adaptive developmental programming
The experience-adaptive hypothesis (Rutter et al., 2004; also referred to as experience-dependent in Greenough et al., 1987, or biological or developmental programming in O’Connor et al., 2003) suggests that during certain sensitive phases of development, an individual adapts to the specific circumstances of their environment, such that they are especially well suited to that environment. This adaptation gets “locked in” and does not change if the child’s environment later changes. Unlike the experience-expectant hypothesis, it is not only aberrant experiences that affect the course of development—an individual will adapt to any experience within a sensitive period.
In some respects, the predictions for the experience-expectant and experience-adaptive hypothesis are similar; both predict the effects of institutionalization to be long lasting and more time in the adoptive home or an older age at assessment shouldn’t substantially lessen the effects. An age at adoption effect is expected, and again it is likely to be a step rather than a continuous function. But in the case of the experience-adaptive hypothesis, there may be variations in outcomes relative to the severity of deprivation experienced in an institution, and the specific outcomes children experience are likely to be adaptive in the institutional environment.
Summary
The above hypotheses all predict that early institutionalization should have a negative impact on development (or, at least make a child ill-adapted for life outside the institution), but they differ in how long the effects are expected to last, and whether specific age periods are important. These hypotheses give direction to our study of the effects of early experience, but lack predictions about specific domains of outcomes, and in fact it is possible that one hypothesis accounts for one domain of outcomes whereas another accounts for other domains of outcomes.
The Current Review
The purpose of this review is to assess the effects of the timing of institution rearing, specifically with respect to social and behavioral outcomes. Age at adoption, a proxy for length of time in the institution, is perhaps the only widely known measure of the institutional experience. To the extent that age at adoption relates to outcomes, it suggests that institutionalization is likely to be a causal factor. Given that the nature of the institutional experience is often not well described, it is particularly crucial to understand the nature of the effect of age at adoption on various outcomes and for various subsamples. While the diversity of studies that address this question and the lack of information that is provided in some studies can hinder definitive conclusions, an analysis of the large number of studies that are available to date serves as a necessary first step to understanding what the current literature suggests about the influence of age at adoption and gives direction to future research that will address this question.
Studies were obtained through searches in PsychInfo (key terms: “age at adoption”, “age at placement”, “length of institution*”, “duration of placement*”, “duration of institution*”) and through references of selected studies. Studies were included if at least half of the sample was likely to have been reared in an institutional setting, and analyses addressed the relation between age at adoption and social or behavioral development. Dissertations were not included in this review. While a meta-analytic approach has its benefits, it is not an ideal strategy for the current review. Many studies lack the information necessary to calculate effect sizes (e.g., they may simply contain a sentence – without any statistics – stating that age at adoption did not relate to the outcome measure). Furthermore, this review seeks to examine whether effects relate to specific outcomes, measurement and analysis strategies, sample characteristics, and the interaction of these variables. In order to adequately address these questions, a meta-analytic approach would often leave only 1–2 studies in each category, whereas the current systematic review approach is not hindered by the heterogeneity in this set of studies.
Studies from the Bucharest Early Intervention Project are included because the age at transition from an institution into a more supportive foster care (not adoptive home) environment was examined. Studies of institutionalized children are included when the age at admission to an institution or duration of institutionalization was examined with respect to social and behavioral outcomes. It is beyond the scope of this review to consider age at adoption effects for intellectual development.
Before age at adoption effects are considered, this set of studies will be reviewed with respect to variables that contribute to our knowledge of a child’s early circumstances, and our ability to attribute children’s outcomes to age at adoption and institutional experience. This section also highlights the degree to which studies provide, or fail to provide, the information that would be necessary in order to draw conclusions about early experience from their findings.
Characteristics of Age at Adoption Studies
Studies that assess outcomes with respect to age at adoption (or, in a few cases, age at admission to an institution) are particularly well-suited to address questions related to the impact of early experience and its timing. But, because of the nature of children’s experience in institutions and the information available to parents and researchers, some conclusions can be limited. For instance, most children enter the institution in the first weeks of life, so it is often impossible to disentangle duration of exposure from specific ages of exposure. Parents may be provided with information about a child’s birthweight and prematurity status, but information about other aspects of a child’s prenatal experience (e.g., exposure to alcohol, tobacco, and poor nutrition), biological family, pre-institutional care, or age at admission to institutional care is typically not available. Further, as a field we typically lack information on a specific child’s quality of care. Parents may be asked to describe the quality of their child’s care in general terms (e.g., “Was your child a ‘favorite’ in the institution?” “Did you suspect your child was neglected?”), but more often the quality of care is generalized from what is typical in a given country.
The set of studies that examines age at adoption effects sometimes lacks information that would allow conclusions to be drawn about the effects of early experience on later development (Table 1). Of 66 studies (31 independent samples, plus 2 meta-analyses), only 8 studies (from 5 independent samples) reported any information about a child’s birthweight, 31 studies (from 10 independent samples) included information about children’s quality of care (of these, 17 studies from 3 independent samples included only a general description, e.g., “the conditions in the institutions were extremely bad”), 23 studies (13 independent samples) addressed whether children had also resided in other early rearing environments, and 23 studies (10 independent samples) reported an approximate age at admission into institutional care for their samples. In 12 studies (4 independent samples), a high correlation between age at adoption and time in institutional care suggests that most children had limited experience in other placements and entered institutional care at an early age. Details about the adoptive family were offered in 40 studies (18 independent samples); in 19 studies (7 independent samples) in which family variables were related to outcomes, findings were mixed, and there was no clear pattern across studies as to which family factors predict which outcomes.
Table 1.
Characteristics of Age at Adoption Studies
| Reference | Prenatal/birth experience | Other placements | Age at admission | Institution-reared | Country of origin | Description of institution | Environment at time of testing | Sample size of target group | ||
|---|---|---|---|---|---|---|---|---|---|---|
| Setting | Specific family characteristics described | Relation to outcomes | ||||||||
| Romania Canada Study | ||||||||||
| Ames (1997) | — | age at adoption & time in institutiona: r (46) = .97, p < .01 | Md = 1m | yes | Romania | specific-globally depriving | adoptive families in Canada | yes | yes | 46 |
| Audet & LeMare (2010) | — | age at adoption & time in institutiona: r (46) = .97, p < .01 | Yes | Romania | — | adoptive families in Canada | yes | yes | 46 | |
| Chisholm (1998) | — | age at adoption & time in institutiona: r (46) = .97, p < .01 | yes | Romania | general-globally depriving | adoptive families in Canada | yes | yes | 46 | |
| Chisholm, Carter, Ames, & Morison (1995) | birthweight | age at adoption & time in institutiona: r (46) = .97, p < .01 | yes | Romania | — | adoptive families in Canada | yes | yes | 46 | |
| Fisher, Ames, Chisholm, & Savoie (1997) | — | age at adoption & time in institutiona: r (46) = .97, p < .001 | M = 1m | yes | Romania | general-globally depriving | adoptive families in Canada | yes | — | 46 |
| English and Romanian Adoptees (ERA) Study | ||||||||||
| Beckett, Bredenkamp, Castle, Groothues, O’Connor, Rutter, and the ERA Study Team (2002) | — | 5 children had 1y + with bio family before institution | — | yes | Romania | specific-globally depriving | adoptive families in UK | — | — | 144 |
| Colvert, Rutter, Beckett, Castle, Groothues, Hawkins, Kreppner, O’Connor, Stevens, & Sonuga-Barke (2008a) | — | — | — | yes | Romania | — | adoptive families in UK | yes | ns | 144 |
| Colvert, Rutter, Kreppner, Beckett, Castle, Groothues, Hawkins, Stevens, & Sonuga-Barke (2008b) | — | — | Usually in neonatal period | yes | Romania | — | Adoptive families in UK | yes | — | 144 |
| Croft, O’Connor, Keaveney, Groothues, Rutter, and the ERA Study Team (2001) | — | age at adoption & time in institutiona: r = .83 | most in first weeks of life | 87% | Romania | general-globally depriving | adoptive families in UK | yes | ns | 165 |
| Kreppner, O’Connor, Rutter, & The ERA Study Team (2001) | — | — | most in first weeks of life | yes | Romania | general-globally depriving | adoptive families in UK | yes | ns | 137 |
| Kreppner, Rutter, Beckett, Castle, Colvert, Groothues, Hawkins, O’Connor, Stevens, & Sonuga-Barke (2007) | birthweight + | age at adoption & time in institutiona: r = .94, p < .001, N = 144 | M = .34m, SD = 1.26 | yes | Romania | general-globally depriving | adoptive families in UK | yes | ns | 144 |
| O’Connor, Bredenkamp, Rutter, & the ERA Study Team (1999) | — | — | 97% by 3rd month | 91% | Romania | General-globally depriving | adoptive families in UK | yes | — | 111 |
| O’Connor, Marvin, Rutter, Olrick, Britner, & The ERA Study Team (2003) | — | — | 85% within first month | yes | Romania | general-globally depriving | adoptive families in UK | — | — | 111 |
| O’Connor, Rutter, & The ERA Study Team (2000) | — | — | 85% within first month | 87% | Romania | general-globally depriving | adoptive families in UK | yes | ns | 165 |
| Rutter, Colvert, Kreppner, Beckett, Castle, Groothues, Hawkins, O’Connor, Stevens, & Sonuga-Barke (2007) | — | age at adoption & time in institutiona: r = .89, p < .001; 80 spent 100%, 53 > 50%, 11 < 50% life in institutions | M = .34m, SD = 1.28m | yes | Romania | general-globally depriving | adoptive families in UK | yes | ns | 144 |
| Rutter, Kreppner, & O’Connor (2001) | — | — | M = 0.34m | 80% most of life | Romania | general-globally depriving | adoptive families in UK | — | — | 165 |
| Rutter, O’Connor, & the ERA Study Team (2004) | — | — | 85% within first month | yes | Romania | general-globally depriving | adoptive families in UK | — | ns | 144 |
| Rutter, Sonuga-Barke, Beckett, Castle, Kreppner, Kumsta, Schlotz, Stevens, & Bell (2010) | birthweight + | 85% whole life in institution; 95.8% at least half life | 85% in first 2 weeks | yes | Romania | general-globally depriving | adoptive families in UK | yes | ns | 144 |
| Sonuga-Barke, Beckett, Kreppner, Castle, Colvert, Stevens, Hawkins, & Rutter (2008) | birthweight | — | — | yes | Romania | general-globally depriving | adoptive families in UK | — | — | 138 |
| Stevens, Sonuga-Barke, Kreppner, Beckett, Castle, Colvert, Groothues, Hawkins, & Rutter (2008) | — | — | most within first 2 weeks | yes | Romania | general-globally depriving | adoptive families in UK | — | — | 144 |
| Bucharest Early Intervention Project (BEIP) | ||||||||||
| Bos, Fox, Zeanah, & Nelson (2009) | birthweight | Many in maternity hospital before institution | 0–17.49m; M = 3.03m, SD = 4.14m | yes | Romania | — | FC (randomized from institution) | — | — | < 93 |
| Smyke, Zeanah, Fox, Nelson, & Guthrie (2010) | — | — | — | yes | Romania | — | FC (randomized from institution) | — | — | 68 |
| Smyke, Zeanah, Gleason, Drury, Fox, Nelson, & Guthrie (2012) | — | — | — | yes | Romania | — | FC (randomized from institution) | — | — | 68 |
| Marcovitch Studies | ||||||||||
| Marcovitch, Cesaroni, Roberts, & Swanson (1995) | — | 56% whole life in institution; others: FC, maternity hospital, other institution, or bio family before institution | 56% at birth | 54.7 %; others adopted from bio families | Romania | genera l-globally depriving | adoptive families in Canada | yes | — | 130 |
| Marcovitch, Goldberg, Gold, Washington, Wasson, Krekewich, & Handle y-Derry (1997) | — | age at adoption & time in institutiona: r = .86 | yes | Romania | general-globally depriving | adoptive families in Canada | yes | — | 56 | |
| Ellis, Fisher, & Zaharie (2004) | — | — | 0–4y; M = 4m | yes | Romania | — | institutions in Romania | n/a | n/a | 51 |
| Other Eastern Europe | ||||||||||
| St. Petersburg-USA Orphan age Research Team (St. Pete Study) | ||||||||||
| Hawk & McCall (2011) | — | age at adoption & time in institutiona: r = .86, p < .001 | yes | mostly Russia | specific-socially-emotionally depriving | adoptive families in USA | — | — | 316 | |
| Merz & McCall (2010) | birthweight + | age at adoption & time in institutiona: r = .77, p < .001 | yes | Russiab | specific-socially-emotionally depriving | adoptive families in USA | yes | — | 342 | |
| Merz & McCall (2011) | birthweight + | age at adoption & time in institutiona: r = .77, p < .001 | yes | mostly Russia | specific-socially-emotionally depriving | adoptive families in USA | yes | — | 418 | |
| Miller, Chan, Tirella, & Perrin (2009) | — | — | — | yesc | Eastern Europe/Former Soviet Union | — | adoptive families in USA | yes | — | 50 |
| Oliveira, Soares, Martins, Silva, Marques, Baptista, & Lyons-Ruth (2012) | + | 26% had no prior placements | 26% at birth, 24% < 5m, 23% 5–12m, 16.2% 12–18m, 10.8% 18–24m | yes | Portugal | — | institutions in Portugal | n/a | n/a | 74 |
| Western Europe | ||||||||||
| Tizard Studies | ||||||||||
| Tizard & Hodges (1978) | all healthy | — | < 4m | yes | UK | specific-adequate | adoptive families in UK | yes | — | 25 |
| Tizard & Rees (1975) | all healthy | — | < 4m | yes | UK | specific-adequate | adoptive families in UK | yes | — | 24 |
| Vorria, Rutter, Pickles, Wolkind, & Hobsbaum (1998a,b)d | — | 18 lived with bio family, 18 lived in another residential unit | nearly half admitted around 5–6y; 16 admitted < 3.5y | yes | Greece | specific-socially-emotionally depriving | institutions in Greece | n/a | n/a | 41 |
| Wolkind (1974)d | — | 58.7% in institution over half life | < 2y, > 2y | yes | UK | — | institutions in UK | n/a | n/a | 92 |
| Middle East | ||||||||||
| Gleitman & Savaya (2011) | + | count of pre-adoption placements | — | about 50% | Israel | specific-adequate | adoptive families in Israel | — | — | 169 |
| Africa | ||||||||||
| Nielsen, Coleman, Guinn, & Robb (2004) | — | — | — | yes | Uganda | specific-quality of care varies by institution | institut ions in Ugand a | n/a | n/a | 33 |
| China | ||||||||||
| Tan Studies | ||||||||||
| Tan (2006) | — | some spent time in FCc | — | yes | Chinae | — | adoptive families in USA | yes | — | 115 |
| Tan & Camras (2011) | — | some spent time in FCc | — | yesc | Chinae | — | adoptive families in USA | yes | — | 869 |
| Tan, Camras, Deng, Zhang, & Lu (2012) | — | some spent time in FCc | — | yesc | Chinae | — | adoptive families in USA | yes | mixed | 133 |
| Tan & Marfo (2006) | — | some spent time in FCc | — | yesc | Chinae | general-socially-emotionally depriving | adoptive families in USA, Canada, Western Europe, and Australia | yes | — | 695 |
| Tan, Marfo, & Dedrick (2007) | — | some spent time in FCc | — | yesc | Chinae (specialneeds adoptions) | specific-socially-emotionally depriving | adoptive families in USA | yes | mixed | 1093 |
| Tan, Marfo, & Dedrick (2010) | — | some spent time in FCc | — | yesc | Chinae | specific-socially-emotionally depriving | adoptive families in USA | yes | mixed | 452 |
| Cohen & Farnia (2011) | — | 18% orphanage care & FC; FC duration: M = 5.91m, SD = 4.11m; 4.3% FC only | — | 77% orphanage care only | Chinae | — | adoptive families in Canada | yes | — | 70 |
| Dalen & Rygold (2006) | — | — | — | yesc | Chinae | — | adoptive families in Norway | — | — | 77 |
| Elliott & McMahon (2011) | — | — | — | 87% | Chinae | — | adoptive families in Australia | yes | — | 59 |
| Rojewski, Shapiro, & Shapiro (2000) | — | — | — | — | Chinae | — | adoptive families in USA | yes | — | 45 |
| Other Asia | ||||||||||
| Goldney, Donald, Sawyer, Kosky, & Priest (1996) | — | — | — | 100%c | Indonesia | — | adoptive families in Australia | — | — | 34 |
| Lee, Seol, Sung, Miller, & the Minnesota International Adoption Project Team (2010)d | — | Some lived with bio family, FC, or in hospital | < 2y, > 2y | yes | South Korea | specific-adequate | institutions in South Korea | n/a | n/a | 612 |
| Multiple Countries | ||||||||||
| Barcons Studies | ||||||||||
| Barcons, Abrines, Brun, Sartini, Fumadó, & Marre (2012) | — | — | — | 86.2% | 28% Asia, 47% Eastern Europe, 14% Latin America, 11% | — | adoptive families in Spain | — | — | 116 |
| Barcons-Castel, Fornieles-Deu, & Costas-Moragas (2011) | — | 14% family care, 14% foster care | — | 92.3% | 51.9% Asia, 26.9% Eastern Europe, 15.4% Central and South America, 5.8% Africa | — | adoptive families in Spain | yes | — | 52 |
| Intl. Adoption Project (IAP) Team | ||||||||||
| Bruce, Tarullo, & Gunnar (2009) | birthweight + | At least 70% of life in institution; < 6m family care before adoption | — | yes | Institution: Eastern Europe, Chinae. FC: South Korea | specific-socially-emotionally depriving | adoptive families in USA | — | — | 40 |
| Gunnar, Van Dulmen, & the International Adoption Project Team (2007) | — | 75% of life in institutions before adoption | — | yes | 29.6% Russia/Eastern Europe, 49.3% Asia, 20.2% Latin America/Carribean | — | adoptive families in USA (Minnesota) | yes | — | 899 |
| Verhulst, Althaus, & Versluis-Den Bieman (1990) | — | — | — | 80.1%c | multiple | — | adoptive families in the Netherlands | — | — | 2148 |
| Verhulst Studies | ||||||||||
| Verhulst, Althaus, & Versluis-Den Bieman (1992) | — | — | — | 80.1%c | 32.0% Korea, 14.6% Colombia, 9.5% India, 7.9% Indonesia, 6.7% Bangladesh, 4.9% Lebanon, 50% Austria, 4.2% other European countries, 15.2% other non-European countries | — | adoptive families in the Netherlands | yes | — | 2148 |
| Lindblad studies | ||||||||||
| Elmund, Lindblad, Vinnerljung, & Hjern (2007) | — | some lived with bio family or FC | — | about 50% | 27% South Asia, 30.7% East Asia, 27% Latin America, 2.3% Africa, 3.3% Middle East | — | adoptive families in Sweden | yes | yes | 16522 |
| Lindblad, Weitoft & Hjern (2010) | — | — | — | about 50%c | Far east, South Asia, Latin America, Eastern Europe, Middle East/Africa | — | adoptive families in Sweden | yes | — | 161 34 |
| Camras, Perlman, Wismer Fries, & Pollack (2006) | — | 15 Chinese children had FC, duration: M = 8.4m | — | yes | Chinae, Eastern Europe | — | adoptive families in USA | — | ns | 41 |
| Cederblad, Höök, Irhammer, & Merke (1999) | — | — | — | some | 36% India, 15% Thailand, 15% Chile, 2% Ethiopia; 32% other Asia or Latin-America | — | adoptive families in Sweden | yes | mixed | 147 |
| Dalen (2001) | — | — | — | yesc | Colomb ia, Korea | — | adoptive families in Norway | — | — | 193 |
| Decker & Omori (2009) | — | — | — | –c | — | — | — | — | — | 141 |
| Habersaat, Tessier, & Pierreh umbert (2011) | — | — | — | yesc | Eastern Europe, East Asia, Latin Americac | — | adoptive families in Canada | — | — | 350 |
| Kvifte Andresen (1992) | — | — | — | — | South Korea, Other | — | adoptive families in Norway | yes | — | 151 |
| Niemann & Weiss (2012) | — | For institution, FC, and bio family: 23% 1 type care, 63% 2 types, 14% 3 types | — | some | 68% Asia, 14% Africa, 14% Central America, 5% Russia/Eastern Europe | — | adoptive families in USA | yes | ns | 22 |
| Meta-analyses Juffer & van Ijzendoorn (2005) | — | — | — | some | multiple | — | adoptive families in multiple countries | — | — | 14581 in 29 studies |
| van den Dries, Juffer, van Ijzendoorn, & Bakermans-Kranenburg (2009) | — | — | — | somef | multiple | — | adoptive families in USA, Canada, Europe, or Australia | — | — | 722 in 17 studies |
“+” = other prenatal information, BEIP = Bucharest Early Intervention Project. FC = Foster Care m = months, y = years, M = mean, Md = median, SD = standard deviation
The high correlations suggest that limited time is left unaccounted for, so children were likely admitted to the institution early in life and experienced limited alternative placements
Data from other countries was also included in this study, but age at adoption analyses were only included for Russian adoptees
This article did not specify whether children were institution-reared; article authors were contacted to obtain this information
These studies examined age at admission into institutional care in addition to or in lieu of examining age at adoption effects
The vast majority of children adopted from China are female
This meta-analysis did not specify whether children were institution-reared, but many of the studies that were analyzed included institution-reared children
In most cases, it is probable that information that was not reported in articles is simply not available, but in other cases, available information is not reported. For example, of the 15 studies of one sample from the ERA Study Team, only 3 included information about children’s prenatal and birth experience. But, with one exception (Bos, Fox, Zeanah, & Nelson, 2009), studies have not found a relationship between birth variables and outcome variables (Kreppner et al., 2007; Merz & McCall, 2010, 2011; Sonuga-Barke et al., 2008). Of the set of 65 studies, 17 did not indicate whether children were reared in an institution; when possible, authors of these studies were contacted for further information, and information was obtained for 12 of these articles. The 32 articles for which no information was available (Decker & Omori, 2009; Kvifte Andresen, 1992; Rojewski et al., 2000) are excluded from further analyses so as to limit the discussion to the effects of early institutional care.
In Table 1, the institutions that were sufficiently described are broadly categorized as globally depriving, socially-emotionally depriving or adequate, roughly in accord with Gunnar’s (2001) guidelines. Globally depriving institutions are those in which the care was deficient in most every respect that was mentioned. Socially-emotionally depriving institutions were deficient in some respects (e.g., routinized care that was insensitive to children’s signals) but health and nutrition needs were met. Adequate institutions were described as mostly good quality, though new caregivers may rotate through quite frequently. Generally, the orphanages in 1990s Romania are described as globally depriving, those in Russia, Greece, and parts of Asia are described as socially-emotionally depriving, and those in parts of Western Europe and South Korea are described as adequate. Some studies assess children who had resided in institutions in a variety of countries as a set. While these studies are included in this review, it is recognized that combining children who were exposed to varying levels of deprivation is bound to blur any effects of age at adoption if age at adoption relates to severity of deprivation.
Age at Adoption
Age at adoption (and age at admission to institutions, when available) studies will be reviewed for attachment and parent-child relationships, disinhibited social behavior, theory of mind and quasi-autistic patterns, peer relationships, social competence and social problems, behavior problems, and attention problems. If outcomes relate to the age at which children enter or leave an institution, it most likely indicates that institutionalization has an effect on development, and the specific ages or duration of exposure to institutionalization is important. Within each category of outcomes, results will be addressed with respect to severity of deprivation, age at assessment, methodological issues, the form of the relationship between age at adoption and outcomes, and whether any associated variables relate to or account for the age at adoption effect.
Attachment & Parent-Child Relationships
Of the 19 studies that examined age at adoption effects for attachment and parent-child relationships, 6 (3 independent samples and 1 meta-analysis) found a significant effect, and in each case, later-adopted children had less favorable outcomes; 4 additional studies (3 independent samples) found effects for only some analyses (Table 2).
Table 2.
Attachment & Parent-Child Relationships
| Study | Country of origin | Severity of deprivation | Age at assessment | N | Construct measured | Outcome measure | Age at adoption: Distribution | Age at adoption: Measurement | Statistical test | Significant effect of age at adoption? | Form of the relationship? | Relationship with other variables? |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Positive Results | ||||||||||||
| Cederblad, Höök, Irhammer, & Merke (1999) | Multiple | — | 13–27y | 147 | presence or absence of attachment with parent or sibs | parent interview | 54% < 7m; 30% > 1y | < 1y, > 1y at adoption and < 6m or > 6m in institution | — | yes, but no statistical test | — | — |
| Gleitman & Savaya (2011) | Israel | Adequate | 11–19.5y | 169 | communication with parents | Family/Parental Relationships Family/Parental Relationships module of Student Questionnaire (self-report) | 0–9y; 57.4% < 2y, 42.6 2–9.5y | Continuous | correlation | yes | — | — |
| Tan & Marfo (2006) | China1 | Socially-emotionally Socially emotionally depriving | 1.5y–11y (retrospective) | 695 | initial rejecting behaviors (early relationship with adoptive mother) | parent interview | preschool: M = 14.2m, SD = 8.3m; school-age: M = 18.5m, SD = 23.7m | continuous, log transformed | correlation | yes | — | — |
| Tan, Marfo, & Dedrick (2007) | China1 | Socially-emotionally depriving | 1.5y–5y (retrospective) | SN = 63 Non-SN = 689 | initiala daptation to adoption (early relationship with adoptive mother) | parent interview | SN: M = 25.85, SD = 10.82; Non-SN: M = 12.71, SD = 4.87 | continuous | multiple regression | yes | — | — |
| China1 | Socially-emotionally depriving | 6y–15.7y (retrospective) | SN = 61 Non-SN = 280 | initial adaptation to adoption (early relationship with adoptive mother) | parent interview | SN: M = 46.98, SD = 37.45; Non-SN: M = 15.61, SD = 15.80 | continuous | multiple regression | yes | — | — | |
| Tan, Marfo, & Dedrick (2010) | China1 | Socially-emotionally depriving | 1.5y–5y (retrospective) | 452 | initial adaptation to adoption (early relationship with adoptive mother) | parent interview | 3–50m; M = 13.1m, SD = 5.1m | continuous, natural log transformed | correlation | yes | — | — |
| van den Dries, Juffer, van Ijzendoorn, & Bakermans-Kranenburg (2009) | Multiple | — | all ages | 17 studies, 722 participants | attachment security | various measures | — | < 12m, > 12m | Q (compared effect sizes) | yes | — | — |
| Mixed Results | ||||||||||||
| Croft, O’Connor, Keaveney, Groothues, Rutter, & the ERA Study Team (2001) | Romania | Globally depriving | 4y, 6y | 165 | parent-child relationship | coded observations: Parent Positivity, Parent Negativity, Child Positivity, Child Negativity | 52 < 6m, 586–24m, 4824–42m | continuous and < 6m, 6–24m, > 24m | correlation; multiple regression | yes; effect disappears when later-adoptees (> 24m) are removed from analyses | — | mediated by cognitive ability |
| O’Connor, Marvin, Rutter, Olrick, Britner, & the ERA Study Team (2003) | Romania | Globally depriving | 4y | 111 | attachment security | separation and reunion procedure | 58 < 6m, 597–24m | continuous and < 6m, 7–24m | chi-squared | no | — | — |
| Romania | Globally depriving | 4y | 111 | attachment organization2 | separation and reunion procedure | 58 < 6m, 597–24m | continuous and < 6m, 7–24m | chi-squared | yes | — | — | |
| Habersaat, Tessier, & Pierrehumbert (2011) | Multiple | — | 12–18y | 350 | attachment security | parent questionnaire (measure not named) | 83 < 6m, 526–12m, 4312–24m, 172 > 24m | < 6m, 6–12m, 12–24m, > 24m | — | yes, for < 6m and > 24m; no for < 12m and > 12m | — | — |
| Smyke, Zeanah, Fox, Nelson, & Guthrie (2010) | Romania | Globally depriving3 | 3.5y | 68 | attachment security | separation and reunion procedure | > 6m4 | dichotimized at various points (18m, 20m, 22m, 24m, 26m, 28m) | chi-squared | yes; only for cut-offs of 24m and later | deflection point (24m) evident | — |
| Romania | Globally depriving3 | 3.5y | 68 | attachment organization2 | separation and reunion procedure | > 6m4 | dichotimized at various points (18m, 20m, 22m, 24m, 26m, 28m) | chi-squared | yes (at all age at adoption cutoffs) | possibly linear | — | |
| Null Results | ||||||||||||
| Ames (1997) | Romania | Globally depriving | most 4.5y; some 5.5–9y | 46 | attachment security | separation and reunion procedure | 8–68m; Md = 18.5m5 | continuous5 | — | no | — | — |
| Chisholm, Carter, Ames, & Morison (1995) | Romania | Globally depriving3 | 1y5m–6y4m | 46 | attachment security | 23-item Attachment Q-sort (parent-report) | 8–68m; Md = 18.5m | continuous | correlation | no | — | — |
| Romania | Globally depriving3 | 1y5m–6y4m | 46 | parent’s attachment to child | Parenting Stress Index parent attachment subscale (parent-report) | 8–68m; Md = 18.5m | continuous | correlation | no | — | — | |
| Barcons, Abrines, Brun, Sartini, Fumadó, & Marre (2012) | Multiple | — | 8–11y | 116 | attachment security | Friends and Family Interview (self-report) | 1–103m; M = 30.61m, SD = 21.94m | < 12m, 13–36m, > 36m | chi-squared | no | — | — |
| Multiple | — | 8–11y | 116 | relationship with parents | BASC subscale (self-report) | 1–103m; M = 30.61m, SD = 21.94m | continuous | linear regression | no | — | — | |
| Cohen & Farnia (2011) | China1 | — | 8m–4y | 70 | attachment security | 23-item Attachment Q-sort (parent-report) | 18–21m; M = 13m | continuous | regression | no | — | — |
| Marcovitch, Goldberg, Gold, Washington, Wasson, Krekewich, & Handley-Derry (1997) | Romania | Globally depriving | 3–5y | 56 | attachment organization2 | separation and reunion procedure | 0–48m; 37 < 6m, 19 > 6m | — | — | no | — | — |
| Nielsen, Coleman, Guinn, & Robb (2004) | Ug and a | Various levels of deprivation | 1–3y | 33 | child’s need for physical contact with adult | observation of free play | 3m–3y; M = 1.5y, SD = .67y6 | < 1y, 1–2y, 2–3y6 | MANCOVA | no | — | — |
| Niemann & Weiss (2012) | Multiple | — | 1–2y | 22 | attachment security | 90-item Attachment Q-sort (experimenter tested) | 6–19m; M = 12.27m | continuous | correlation | no | — | — |
| Tan (2006) | China1 | Socially-emotionally depriving2 | 6–8y (retrospective) | 115 | initial rejecting behaviors (early relationship with adoptive mother) | parent interview | < 2y; M = about 9m, SD = about 4.5m | continuous | multiple regression | no | — | — |
| Tizard & Hodges (1978) | UK | Adequate | 8y | 25 | attachment security | parent interview | most 2–4y; 54.5–7.5y | — | — | no | — | — |
Note. Top section includes positive results, middle section includes mixed results, bottom section indicates null results
SN = special needs, m = months, y = years, M = mean, Md = median, SD = standard deviation, BASC = Behavior Assessment System for Children
The vast majority of children adopted from China are female
Attachment Organization refers to the distinction between MacArthur’s categories of secure (B), avoidant (A), and ambivalent-dependent (C), vs. disorganized-controlling (D) and insecure-other (I–O)
This article did not provide information about the conditions of these institutions, but other studies from the same authors on the same sample did describe the severity of the deprivation.
In this study, children moved from an institution to foster care (not an adoptive home) at this age
This study used time in the orphanage instead of age at adoption; these two variables were correlated at r = .97
This study assessed children who were still institutionalized, so duration of institutionalization was assessed rather than age at adoption
Severity of deprivation
Studies that found effects were more likely to be less severely depriving; however, this set of studies had substantially larger sample sizes than most studies that failed to detect an effect of age at adoption, so conclusions with regard to severity of deprivation cannot be made.
Age at assessment
Age at assessment does not appear to have an appreciable impact on whether studies found effects in this set of studies.
Methodological issues
Nearly every study that contained over 100 participants found a significant effect of age at adoption while nearly every study that included fewer participants failed to find an effect; in Tan’s studies (Tan, 2006; Tan & Marfo, 2006; Tan et al., 2007, 2010) of relatively less depriving Chinese institutions, it is possible that the age at adoption effect was less pronounced such that only studies with over 300 participants had enough power to detect the effect.
Studies that used parent interviews were most likely to detect effects. In two samples (Habersaat et al., 2011; Smyke et al., 2010), age at adoption effects on attachment security are most evident when children adopted after 24 months of age are compared with those adopted earlier. Age at adoption effects for attachment organization, on the other hand, do not appear to be dependent on the specific age at adoption cut-offs that are utilized (Smyke et al., 2010). There was no apparent effect of the specific analyses used to examine the age at adoption effect, or the distribution of age at adoption.
Form of the relationship
While evidence of the form of the relationship between age at adoption and attachment outcomes is limited, one study (Smyke et al., 2010) suggests a possible linear relationship for attachment organization, but a deflection point at 24 months at adoption is evident for attachment security. An additional study from a different Romanian sample (Croft et al., 2001) noted that an age at adoption effect for the parent-child relationship was no longer significant once children adopted after 24 months of age were excluded.
Associated variables
One study (Croft et al., 2001) found that the effect of age at adoption on the parent-child relationship was mediated by cognitive ability.
Summary
Generally, later-adopted children have more difficulty with attachment and parent-child relationships, and effects were more evident in larger samples. Effects were also most evident through parent interviews, suggesting that while parent observation picks up on differences between earlier- and later-adopted children, these differences are not always apparent on standardized measures. The age at adoption effect may be linear for attachment organization and attachment security may be worse when children are adopted after 24 months of age from globally depriving Romanian institutions. Cognitive ability might serve as a mediator for age at adoption effects on the parent-child relationship.
Reactive Attachment Disorder (RAD) and Disinhibited Social Behavior (DSB)
Of the 15 studies (6 independent samples) that investigated the relationship between age at adoption and RAD or DSB (alternately called disinhibited attachment, indiscriminate friendliness, or over-friendliness), 7 (2 independent samples) found significant effects (Table 3); in each case, children adopted at a later age were more likely to demonstrate disinhibited social behavior. Two additional studies detected effects in some analyses, but not others. One study of institutionalized children in Portugal failed to detect effects for age at admission to the institution.
Table 3.
Reactive Attachment Disorder and Disinhibited Social Behavior
| Study | Country of origin | Severity of deprivation | Age at assessment | N | Construct measured | Outcome measure | Age at adoption: Distribution | Age at adoption: Measurement | Statistical test | Significant effect of age at adoption? | Form of the relationship? | Relationship with other variables? |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Positive Results | ||||||||||||
| Bruce, Tarullo, & Gunnar (2009) | Eastern Europe, China1 | Socially-emotionally depriving | 6–7y | 40 | disinhibited social behavior | observational measure; parent interview2 | < 36m | continuous | correlation | yes | — | mediated by inhibitory control |
| O’Connor, Bredenkamp, Rutter, & The ERA Study Team (1999) | Romania | Globally depriving | 4y | 111 | disinhibited or inhibited attachment behavior | 3-item parent interview3; additional items for inhibited attachment | < 24m | continuous and < 6m, 7–24m | correlation | yes | linear | not mediated by nutritional deprivation or cognitive impairment |
| O’Connor, Rutter, & The ERA Study Team (2000) | Romania | Globally depriving | 4y, 6y | 165 | disinhibited or inhibited attachment behavior | 3-item parent interview3; additional items for inhibited attachment | 58 < 6m, 59 6–24m, 48 > 24m | continuous and < 6m, 6–24m, > 24m | correlation, chi-squared | yes | linear | not accounted for by other behavioral or emotional problems, or by cognitive ability; stable from 4y to 6y |
| Rutter, Kreppner, & O’Connor (2001) | Romania | Globally depriving | 6y | 165 | disinhibited or inhibited attachment behavior | 3-item parent interview3; additional items for inhibited attachment | 58 < 6m, 59 6–24m, 48 > 24m | < 6m, 6–24m, > 24m | chi-square trends | yes | — | — |
| Rutter, O’Connor, and the ERA Study Team (2004) | Romania | Globally depriving | 4y, 6y | 144 | disinhibited attachment | 3-item parent interview3 | 45 < 6m, 54 6–24m, 45 > 24m | < 6m, 6–24m, > 24m | chi-square; correlation | yes | — | — |
| Rutter, Sonuga-Barke, Beckett, Castle, Kreppner, Kumsta, Schlotz, Stevens, & Bell (2010) | Romania | Globally depriving | 6y, 11y, 15y | 144 | disinhibited attachment | 3-item parent interview3 | 58 < 6m, 59 6–24m, 48 > 24m | < 6m (combined with non-PI), > 6m | Fisher’s exact test | yes | — | partial mediation by head circumference (20% of total effect) |
| Sonuga-Barke, Beckett, Kreppner, Castle, Colvert, Stevens, Hawkins, & Rutter (2008) | Romania | Globally depriving | 11y | 138 | disinhibited attachment | 3-item parent interview3 | .5–3y7m | < 6m, > 6m | ANOVA; regression | yes | — | overall sample: not mediated by head circumference; sub-nourished subgroup: possibly mediated by head circumference |
| Mixed Results | ||||||||||||
| Rutter, Colvert, Kreppner, Beckett, Castle, Groothues, Hawkins, O’Connor, Stevens, & Sonuga-Barke (2007) | Romania | Globally depriving | 4y, 6y, 11y | 111 | disinhibited attachment | 3-item parent interview3 | 44 < 6m, 88 6–42m | continuous and < 6m, 6–42m | chi-square; phicoefficient (contrast); t-test | yes, for persistence of DA over time, and for correlation among 6–42m at adoption group, but ns for correlation of whole sample | — | — |
| Smyke, Zeanah, Gleason, Drury, Fox, Nelson, & Guthrie (2012) | Romania | Globally depriving4 | 6m–8y | 68 | disinhibited attachment disorder | Disturbances of Attachment Interview (first 5 items) 5 (parent-report) | Md = 24m6 | < 24m, > 24m6 | repeated-measures ANOVA | yes | — | interaction with age at assessment |
| Romania | Globally depriving4 | 6m–8y | 68 | inhibited attachment disorder | Disturbances of Attachment Interview (last 3 items) 7 (parent-report) | Md = 24m6 | < 24m, > 24m6 | repeated-measures ANOVA | no | — | — | |
| Null Results | ||||||||||||
| Ames (1997) | Romania | Globally depriving | most 4.5y; some 5.5–9y | 46 | indiscriminate friendliness | 5-item parent interview8 | 8–68m; Md = 18.5m9 | continuous9 | — | no | — | — |
| Chisholm (1998) | Romania | Globally depriving | 1y5m–9y2m | 46 | indiscriminate friendliness | 5-item parent interview8 | 8–68m; Md = 18.5m9 | —9 | — | no | — | — |
| Chisholm, Carter, Ames, & Morison (1995) | Romania | Globally depriving4 | 1y5m–6y4m | 46 | indiscriminate friendliness | 5-item parent interview8 | 8–68m; Md = 18.5m | continuous | correlation | no | — | — |
| Nielsen, Coleman, Guinn, & Robb (2004) | Ug and a | Various levels of deprivation | 1–3y | 33 | discriminating social behavior | observation of free play | 3m–3y; M = 1.5y, SD = .67y10 | < 1y, 1–2y, 2–3y10 | ANCOVA | no | — | — |
| Tizard & Hodges (1978) | UK | Adequate | 8y | 25 | “over-friendliness” | parent interview or observations | 2y–7.5y | — | — | no | — | — |
| Tizard & Rees (1975) | UK | Adequate | 4.5y | 24 | “over-friendliness” or “indiscriminately affectionate” | observations | M = 3.11y; SD = .69 | — | — | no | — | — |
| Age at admission: Distribution | Age at admission: Measurement | Significant effect of age at admission | ||||||||||
| Oliveira, Soares, Martins, Silva, Marques, Baptista, & Lyons-Ruth (2012) | Portugal | — | 11–30m | 74 | indiscriminate behavior | RISE (observational measure based on SSP) | 26% at birth, 24% < 5m, 23% 5–12m, 16.2% 12–18m, 10.8% 18–24m | continuous and various cutoff points | correlation, other tests not listed | no | — | — |
DA = disinhibited attachment, RISE = Rating of Infant and Stranger Engagement, SSP = Strange Situation Procedure, m = months, y = years, M = mean, Md = median, SD = standard deviation
The vast majority of children adopted from China are female
Items were: child too eager to approach, made personal comments to adults, initiated physical contact with unfamiliar adults
Items were: definite lack of differentiation among adults with respect to the child’s social response to them, clear indication that the child would readily go off with a stranger, definite lack of checking back with parent in anxiety-provoking situations
This article did not provide information about the conditions of these institutions, but other studies from the same authors on the same sample did describe the severity of the deprivation.
Items were: whether the child has developed a preference for a specific caregiver, approaches the caregiver for comfort, responds to comfort when offered, engages in reciprocal social interaction, regulates emotions well
In this study, children moved from an institution to foster care (not an adoptive home) at this age
Items were: whether the child checks back with the caregiver when exploring, shows age-appropriate reticence around strangers, demonstrates willingess to “go off” with a stranger
Items were: child wandered without distress, child would be willing to go home with a stranger, child was friendly to strangers, child was ever shy, what a child typically did upon meeting new adults
This study used time in the orphanage instead of age at adoption; these two variables were correlated at r = .97
Severity of deprivation
Of two independent samples (ERA Study and Canada Study) of children adopted from globally depriving Romanian institutions, only one (ERA Study) detected significant effects of age at adoption in at least some of the analyses. An additional study of children adopted from socially-emotionally depriving institutions (Bruce et al., 2009) detected an age at adoption effect, while studies of children who experienced various levels of deprivation and adequate institutions failed to detect age at adoption effects.
Age at assessment
Age at assessment appears to have a limited effect, at best. In the ERA Study, disinhibited attachment may persist from age 4 to 6 years (O’Connor, Rutter, & The ERA Study Team, 2000; Rutter et al., 2004) and from age 6 to 11 years for children adopted after 6 months of age (Rutter et al., 2007). One BEIP study showed that children adopted after 24 months of age had higher levels of disinhibited RAD than earlier-adopted children at ages 30 and 54 months of age (the difference was nonsignificant at Baseline, 42 months, and 8 years; Smyke et al., 2012).
Methodological issues
Age at adoption effects were more likely to be detected with parent interviews than with observational measures, and when assessed with a 3-item parent interview (ERA Study) than with a 5-item parent interview (Canada Study). Because the 3-item and 5-item parent interviews are quite similar in content (see footnotes in Table 3), it is doubtful that the nature of the measures is responsible for whether studies find effects. One study found that while a correlation between age at adoption and disinhibited attachment was nonsignificant in the whole sample, it reached significance when earlier-adopted (< 6 months of age at adoption) children were excluded (Rutter et al., 2007). Furthermore, the BEIP demonstrated that age at adoption is related to the disinhibited, but not inhibited, type of RAD (Smyke et al., 2012). It is unclear whether the statistical tests used influenced whether studies found effects, largely because most studies with null results failed to specify the statistical tests that were used. With one exception (Bruce et al., 2009), each study that had fewer than 100 participants failed to find an effect of age at adoption, whereas each study that had more participants detected an age at adoption effect for disinhibited social behavior.
Form of the relationship
Two ERA Study papers (O’Connor, Bredenkamp, Rutter, & The ERA Study Team, 1999; O’Connor et al., 2000) suggest that there may be a linear relationship between age at adoption and DSB. In another paper by the same group, an effect was only evident when earlier-adoptees (< 6 months of age at adoption) were excluded, suggesting that there is meaningful variability in disinhibited attachment from 6 to 42 months of age at adoption (Rutter et al., 2007).
Associated variables
Inhibitory control may mediate the relationship between age at adoption and disinhibited social behavior (Bruce et al., 2009). The possible role of head circumference (which may be a proxy for brain growth) as a mediator of the age at adoption effect on disinhibited social behavior was examined in several studies from the ERA Study Team. One study detected partial mediation, accounting for 20% of the total effect (Rutter et al., 2010), and another detected mediation only for children who had evidence of sub-nourishment (a small effect; Sonuga-Barke et al., 2008). Additional studies from the same group suggest that the age at adoption effect is not accounted for by nutritional deprivation, other behavioral or emotional problems, or cognitive ability (O’Connor et al., 1999, 2000).
Summary
Two of six samples that examined age at adoption effects for disinhibited social behavior found a significant effect, with later-adopted children faring more poorly. Most studies that detected an age at adoption effect on DSB were studies from the ERA Study team using parent interviews with over 100 participants adopted from globally depriving Romanian institutions. While evidence is limited, data from the ERA study suggests that the age at adoption effect may be linear, and it may be most evident when the earliest-adopted children (< 6 months of age at adoption) are excluded. The age at adoption effect on disinhibited social behavior may be mediated by inhibitory control; the potential mediation via head circumference is less clear. The age at adoption effect is not mediated by nutritional deprivation, other behavioral or emotional problems, or cognitive ability.
Quasi-Autism and Theory of Mind
The age at adoption effect for quasi-autistic features or theory of mind were examined in four studies, all from the ERA Study Team; the studies were consistent in finding that later-adopted children are more likely to have problems in these domains (Table 4). One study found that quasi-autism was present in 0.9% of children in a pooled comparison group (UK adoptees, non-PI Romanian adoptees, PI Romanian adoptees adopted before 6 months of age), and in 14.9% of PI children adopted after 6 months of age (Rutter et al., 2010).
Table 4.
Quasi-Autism and Theory of Mind
| Study | Country of origin | Severity of deprivation | Age at assessment | N | Construct measured | Outcome measure | Age at adoption: Distribution | Age at adoption: Measurement | Statistical test | Significant effect of age at adoption? | Form of the relationship? | Relationship with other variables? |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Positive Results | ||||||||||||
| Rutter, Kreppner, & O’Connor (2001) | Romania | Globally depriving | 6y | 165 | Quasiautistic features | ASQ, ADI-R (parent-report, experimenter tested) | 58 < 6m, 59 6–24m, 48 24–42m | < 6m, 6–24m, > 24m | chi square trends | yes | — | — |
| Rutter, Sonuga-Barke, Beckett, Castle, Kreppner, Kumsta, Schlotz, Stevens, & Bell (2010) | Romania | Globally depriving | 6y, 11y, 15y | 144 | Quasiautism | SCQ for screening, then ADI-R and ADOS (parent-report, experimenter tested) | 58 < 6m, 59 6–24m, 48 24–42m | < 6m, > 6m | none (listed % with problems in each group); regression | yes | — | not mediated by head circumference |
| Sonuga-Barke, Beckett, Kreppner, Castle, Colvert, Stevens, Hawkins, & Rutter (2008) | Romania | Globally depriving | 11y | 138 | Quasiautistic features | SCQ (parent-report) | .5m–3y7m | < 6m, > 6m | ANOVA | yes | — | Effect was independent of subnutrition status; not mediated by head circumference |
| Mixed Results | ||||||||||||
| Colvert, Rutter, Kreppner, Beckett, Castle, Groothues, Hawkins, Stevens, & Sonuga-Barke (2008b) | Romania | Globally depriving1 | 11y | 144 | Theory of mind | Strange Stories Task (experimenter tested) | 58 < 6m, 59 6–24m, 48 24–42m | continuous, < 6m, > 6m | correlation, ANOVA, ANCOVA | yes; ns when correlated only within > 6m group | probable step function at 6m | still significant after co-varying IQ |
Note. Top section includes positive results, bottom section includes mixed results
ASQ = Autism Screening Questionnaire, ADI-R = Autism Diagnostic Interview – Revised, SCQ = Social Communication Questionnaire, ADOS = Autism Diagnostic Observation Schedule, m = months, y = years
This article did not provide information about the conditions of these institutions, but other studies from the same authors on the same sample did describe the severity of the deprivation.
Severity of deprivation, age at assessment, and methodological issues
Because each study utilized the same Romanian sample, and each study found a significant effect, it is impossible to assess whether severity of deprivation, age at assessment, or methodological issues play a role in the pattern of findings.
Form of the relationship
In one study (Colvert et al., 2008b), children adopted before 6 months of age had fewer problems than later-adopted children, but variation in age at adoption after 6 months does not appear to be meaningful. While consistent with a step function, the authors did not explicitly discuss the form of the age at adoption function.
Associated variables
The age at adoption effect remained significant when IQ was entered as a covariate (Colvert et al., 2008b). Head circumference did not mediate the age at adoption effect for quasi-autistic features, and the effect was also independent the effect of subnutrition (as indexed by weight; Rutter et al., 2010; Sonuga-Barke et al., 2008).
Summary
Only one sample of children addressed the question of whether age at adoption relates to quasi-autistic features or theory of mind, but it is clear that, at least for children from globally depriving institutions, children adopted after 6 months of age are at elevated risk for quasi-autistic features. The findings are consistent with a step function at 6 months of age at adoption, and the effects are not accounted for by head circumference, IQ, or nutrition.
Peer Relationships
Of the 6 studies (4 independent samples) that examined age at adoption effects for peer relationships, 4 found peer relationships to be poorer for later-adopted children, and one found mixed results (Table 5). Of the 3 ERA Study team articles that addressed this topic, 2 detected age at adoption effects; it was not readily apparent what accounted for one study’s failure to detect an effect. One study found null results for age at admission effects.
Table 5.
Peer Relationships
| Study | Country of origin | Severity of deprivation | Age at assessment | N | Construct measured | Outcome measure | Age at adoption: Distribution | Age at adoption: Measurement | Statistical test | Significant effect of age at adoption? | Form of the relationship? | Relationship with other variables? |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Positive Results | ||||||||||||
| Barcons, Abrines, Brun, Sartini, Fumadó, & Marre (2012) | Multiple | — | 8–11y | 116 | social stress | BASC subscale (self-report) | 1–103m; M = 30.61m, SD = 21.94m | continuous | linear regression | yes | — | — |
| Multiple | — | 8–11y | 116 | interpersonal relations with peers | BASC subscale (self-report) | 1–103m; M = 30.61m, SD = 21.94m | continuous | linear regression | yes | — | — | |
| Cederblad, Höök, Irhammer, & Merke (1999) | Multiple | — | 13–27y | 147 | relations to others | Self-esteem questionnaire ‘I think I am’ (self-report) | 54% < 7m; 30% > 1y | < 6m, > 6m; | — | yes | — | — |
| O’Connor, Rutter, & The ERA Study Team (2000) | Romania | Globally depriving | 6y | 165 | peer relationship disturbance | parent interview | 58 < 6m, 59 6–24m, 48 > 24m | continuous | correlation | yes | — | mediated by attachment disturbance |
| Rutter, Sonuga-Barke, Beckett, Castle, Kreppner, Kumsta, Schlotz, Stevens, & Bell (2010) | Romania | Globally depriving | 6y, 11y, 15y | 144 | peer difficulties | Rutter/SDQ subscales (parent & teacher-report) | 58 < 6m, 59 6–24m, 48 > 24m | < 6m, > 6m | chisquare; ANOVA | yes, for parent and teacher report | — | significant at all ages at assessment, but significant interaction with age at assessment: problems increase from 6y to 11y, no further increase from 11y to 15y; evidence of mediation by DSP |
| Mixed Results | ||||||||||||
| Nielsen, Coleman, Guinn, & Robb (2004) | Ug and a | Various levels of deprivation | 1–3y | 33 | time spent with peers, adults, or alone | observation of free play | 3m–3y; M = 1.5y, SD = .67y1 | < 1y, 1–2y, 2–3y1 | MANCOVA | yes: only for interaction | — | interaction between duration of institutionalization and contact with relatives; spent more time with another child if < 1y in institution and contact with relatives, or 2–3y in institution and no contact with relatives |
| Null Results | ||||||||||||
| Rutter Kreppner, & O’Connor (2001) | Romania | Globally depriving | 6y | 144 | peer difficulties | Rutter parent and teacher scales; parent interview | 58 < 6m, 59 6–24m, 48 > 24m | < 6m, 6–24m, > 24m | chisquare trends | no | — | — |
| Age at admission: Distribution | Age at admission: Measurement | Significant effect of age at admission? | — | — | ||||||||
| Vorria, Wolkind, Rutter, Pickles, & Hobsbaum (1998b) | Greece | Socially-emotionally depriving | 9–11y | 41 | peer relationships | parent interview | half admitted around 5–6y, 16 admitted before 3.5y | < 3.5y, 3.5y–5.5y, > 5.5y | — | no | — | — |
Note. Age at adoption findings are listed above age at admission findings. Within age at adoption section, top section includes positive results, middle section includes mixed results, and bottom section includes null results.
BASC = Behavior Assessment System for Children SDQ = Strengths and Difficulties Questionnaire, DSP = Deprivation-specific psychological patterns, m = months, y = years
This study assessed children who were still institutionalized, so duration of institutionalization was assessed rather than age at adoption
Severity of deprivation
There was not sufficient variability in severity of deprivation in these studies to determine whether this was a meaningful factor.
Age at assessment
One study found that while age at adoption effects on peer relationships were detected at each age (6, 11, and 15 years), there was a significant interaction between age at adoption and age at assessment (Rutter et al., 2010). Peer relationship problems significantly increased from age 6 to 11 years, with no further increase from age 11 to 15 years.
Methodological issues
The measures and statistical analyses do not clearly relate to whether studies found an age at adoption effect, but self-report measures may be somewhat more likely to detect effects. One study failed to detect an effect of age at admission into the institution using cut-offs of 3.5 years and 5.5 years (Vorria, Rutter, Pickles, Wolkind, & Hobsbaum, 1998b). It is possible that the null results were due to focusing exclusively on ages beyond the first two years of life.
Form of the relationship
The form of the age at adoption effect was not assessed.
Associated variables
The age at adoption effect on peer relationships may be mediated by attachment disturbance (O’Connor et al., 2000) or broader deprivation-specific problems (DSP; Rutter et al., 2010). In one study from Uganda, institutionalized children spent more time with another child if they were institutionalized less than 1 year with visits from relatives, or if they were institutionalized 2–3 years without family contact (Nielsen, Coleman, Guinn, & Robb, 2004).
Summary
Most studies detected an effect of age at adoption on peer relationships, with later-adopted children faring more poorly. Age at adoption effects may be more prominent among adolescent than childhood samples, and when self-report measures are used. It may be that peer problems become more evident as children enter the more socially complex adolescent years, and because much peer interaction occurs away from parents and teachers, PI children may be the best reporters on their own peer functioning. Attachment and DSP may be mediators of the age at adoption effect on peer relationships, and contact with relatives may moderate the effect of duration of institutionalization for institutionalized children. Age at admission effects were not detected, but this might be explained by the relatively later cut-off ages that were utilized.
Social Competence and Social Problems
Age at adoption effects on social competence and social problems were assessed in 13 studies (11 independent samples); 6 studies (5 independent samples) found that children adopted at relatively older ages have greater difficulties, and 3 studies (3 independent samples) found age at adoption effects only in some analyses (Table 6).
Table 6.
Social Competence and Social Problems
| Study | Country of origin | Severity of deprivation | Age at assessment | N | Construct measured | Outcome measure | Age at adoption: Distribution | Age at adoption: Measurement | Statistical test | Significant effect of age at adoption? | Form of the relationship? | Relationship with other variables? |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Positive Results | ||||||||||||
| Ames (1997) | Romania | Globally depriving | most 4.5y; some 5.5–9y | 46 | adaptive and personal-social problems | Gesell test (experimenter tested) | 8–68m; Md = 18.5m1 | continuous1 | correlation | yes | — | — |
| Romania | Globally depriving | most 4.5y; some 5.5–9y | 46 | social behavior problems | Factor derived from SSRS, PBQ, & CBCL (parent-report) | 8–68m; Md = 18.5m1 | continuous1 | correlation | yes | — | — | |
| Camras, Perlman, Wismer Fries, & Pollack (2006) | China2, Eastern Europe | — | 4–5.5y | 41 | emotion identification | Emotion Situation Task; Emotion Identification Task (experimenter tested) | Chinese: M = 14m, SD = 4.9m; Eastern European: M = 19.1m, SD = 11.6m | continuous | correlation | yes | — | — |
| Cederblad, Höök, Irhammer, & Merke (1999) | Multiple | — | 13–27y | 147 | social problems | CBCL subscale (parent-report) | 54% < 7m; 30% > 1y | < 6m, > 6m; | ANOVA | yes | — | — |
| Gunnar, van Dulmen, & the International Adoption Project Team (2007) | Multiple | — | 4–18y | 899 | social problems | CBCL subscale (parent-report) | M = 29.5m, SD = 30.7m | < 24m, > 24m | logistic regression | yes | — | — |
| Hawk & McCall (2011) | mostly Russia | Socially-emotionally depriving | 6–18y | 316 | social problems | CBCL subscale (parent-report) | 3.3–57.7m; M = 16m, SD = 11.4m | < 18m, > 18m | ANOVA, MANOVA, not specified | yes | step function at 18m | — |
| Merz & McCall (2010) | mostly Russia | Socially-emotionally depriving | 6–18y | 342 | social problems | CBCL subscale (parent-report) | M = 16.4m, SD = 11.75m | < 9, 9–17m, ≥ 18m; < 18m, > 18m | chi square; hierarchical multiple regression | yes | probable step function at 18m | effect is especially prominent for older-at-assessment group (12–18y) |
| Mixed Results | ||||||||||||
| Habersaat, Tessier, & Pierrehumbert (2011) | Multiple | — | 12–18y | 350 | social problems | CBCL subscale (parent-report) | 83 < 6m, 52 6–12m, 43 12–24m, 172 > 24m | < 6m, 6–12m, 12–24m, > 24m | — | yes, for < 6m and > 24m; no for < 12m and > 12m | — | — |
| Nielsen, Coleman, Guinn, & Robb (2004) | Ug and a | Various levels of deprivation | 1–3y | 33 | range of facial expressions; affect variability | observation of free play | 3m–3y; M = 1.5y, SD = .67y3 | < 1y, 1–2y, 2–3y3 | MANCOVA | yes: only for interaction | — | multivariate interaction between duration of institutionalization and contact with relatives; univariate effects were ns |
| Ug and a | Various levels of deprivation | 1–3y | 33 | prosocial behavior | observation of free play | 3m–3y; M = 1.5y, SD = .67y3 | < 1y, 1–2y, 2–3y3 | ANCOVA | no | — | — | |
| Tan & Camras (2011) | China2 | Socially-emotionally depriving4 | M = 6.9y, SD = 2.9y | 869 | social skills | SSRS (parent-report) | M = 15.6m, SD = 13.9m | continuous | regression | Total score: marginal; yes for some subscales | — | — |
| China2 | Socially-emotionally depriving4 | M = 6.9y, SD = 2.9y | 869 | social skills | SSRS (teac herreport) | M = 15.6m, SD = 13.9m | continuous | regression | Total score: yes; yes for some subscales | — | — | |
| Null Results | ||||||||||||
| Barcons-Castel, Fornieles-Deu, & Costas-Moragas (2011) | Multiple | — | 6–11y | 52 | social skills | BASC subscale (parent & self-report) | 8–84m; M = 28.75m, SD = 21.42m | < 12m, 13–36m, > 37m | unifactorial ANOVA | no | — | — |
| Multiple | — | 6–11y | 52 | leadership | BASC subscale (parent & self-report) | 8–84m; M = 28.75m, SD = 21.42m | < 12m, 13–36m, > 37m | unifactorial ANOVA | no | — | — | |
| Dalen & Rygold (2006) | China2 | — | 7–13y | 77 | social skills | SSRS (teacherreport) | 2m–52m; M = 6.88, SD = 9.28 | — | — | no | — | — |
| Tan (2006) | China2 | Socially-emotionally depriving4 | 6–8y | 115 | social probems | CBCL subscale (parent-report) | < 2y | continuous | regression | no | — | — |
| Verhulst, Althaus, & Versluis-Den Bieman (1990) | Multiple | — | 6–21y | 16,134 | social competence | CBCL subscale (parent-report) | 513 0–6m, 221 7–12m, 329 13–24m, 311 25–36m, 278 37–48m, 198 49–60m, 156 61–72m, 142 73–125m | < 6m, 7–12m, 13–24m, 25–36m, 37–48m, 49–60m, 61–72m, 73–125m | ANCOVA | no | — | — |
Note. Top section includes positive results, middle section includes mixed results, and bottom section includes null results.
SSRS = Social Skills Rating System, PBQ = Preschool Behavior Questionnaire, CBCL = Child Behavior Checklist, BASC = Behavioral Assessment System for Children, m = months, y = years, M = mean, Md = median, SD = standard deviation
This study used time in the orphanage instead of age at adoption; these two variables were correlated at r = .97
The vast majority of children adopted from China are female
This study assessed children who were still institutionalized, so duration of institutionalization was assessed rather than age at adoption
These articles did not provide information about the conditions of these institutions, but other studies from the same authors on the same sample did describe the severity of the deprivation
Severity of deprivation
Studies of children adopted from Eastern European globally or socially-emotionally depriving institutions appeared to be most likely to detect an age at adoption effect. Mixed or null effects tended to be found for studies of children adopted from China, though one study of children adopted from China3 or Eastern Europe did detect an effect of age at adoption (Camras et al., 2006).
Age at assessment
One study found that the effect of age at adoption on social problems is especially prominent for adolescents (Merz & McCall, 2010).
Methodological issues
The measures and statistical analyses used, the number of subjects, and distribution and measurement of age at adoption did not appear to have an appreciable impact on the likelihood of detecting an effect. While most studies utilized parent-report, results did not have an apparent relation to whether parents, teachers, or experimenters provided ratings of children’s social behavior.
Form of the relationship
Two studies of one sample of children from socially-emotionally depriving institutions in Russia are consistent with a step function at 18 months of age at adoption (Hawk & McCall, 2011; Merz & McCall, 2010); additional time in the institution beyond 18 months of age does not appear to further increase the risk of social problems.
Associated variables
When range of facial expressions and affect variability were observed in Ugandan institutionalized children, there was a multivariate interaction between duration of institutionalization and visits from relatives; no univariate effects were significant (Nielsen et al., 2004).
Summary
Generally, children adopted at later ages have poorer social competence and more social problems. Effects were more likely to be detected in studies of children adopted from Eastern Europe who were globally or socially-emotionally deprived. There is some evidence of a step function at 18 months of age for children adopted from socially-emotionally depriving Russian institutions.
Behavior Problems
Age at adoption effects on behavior problems were examined in 32 studies (19 independent samples and 1 meta-analysis), and age at admission effects were examined in 3 studies (Table 7). A significant age at adoption effect was detected in 11 studies (7 independent samples), and in each case, later-adopted children had more problems; effects were found in only some analyses or subgroups in 8 additional studies (6 independent samples). A meta-analysis failed to detect an effect, and this may be because studies were grouped as to whether they contained earlier- or later-adopted children and whether PI differed from non-PI children instead of examining the age at adoption effect within studies.
Table 7.
Behavior problems
| Study | Country of origin | Severity of deprivation | Age at assessment | N | Construct measured | Outcome measure | Age at adoption: Distribution | Age at adoption: Measurement | Statistical test | Significant effect of age at adoption? | Form of the relationship? | Relationship with other variables? |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Positive Results | ||||||||||||
| Ames (1997) | Romania | Globally depriving | most 4.5y; some 5.5–9y | 46 | stereo typed behavior problems | parent interview | 8–8m; Md = 18.5m1 | continuous1 | correlation | yes | — | — |
| Romania | Globally depriving | most 4.5y; some 5.5–9y | 46 | behavior problems | CBCL (parent-report) | 8–8m; Md = 18.5m1 | continuous1 | correlation | yes, for Total, Internalizing, and Externalizing | — | — | |
| Fisher, Ames, Chisholm, & Savoie (1997) | Romania | Globally depriving | 18–76m | 46 | behavior problems | CBCL (parent-report) | 806m; M = 18.51 | continuou1 | correlation | yes for Total and Internalizing, marginal for Externalizing | — | — |
| Romania | Globally depriving | 18–76m | 46 | rate of improvement in stereo typed behavior problems | parent interview | 8–6m; M = 18.51 | continuous1 | correlation | yes | — | — | |
| Beckett, Bredenkam, Castle, Groothues, O’Connor, Rutter, & The ERA Study Team (2002) | Romania | Globally depriving | 6y | 144 | Rocking, self-injury, unusual sensory interests, problems with chewing and swallo wing | parent interview | 45 0–6m, 19 6–2m, 35 12–24m, 45 24–42m | < 6m, 6–12m, 12–24m, 24–42m | chi square trends, ANOVA | yes | linear, step function at 6m | — |
| Rutter, Sonuga-Barke, Beckett, Castle, Kreppner, Kumsta, Schlotz, Stevens, & Bell (2010) | Romania | Globally depriving | 6y, 11y, 15y | 144 | Deprivation Specific Psychological Patterns | most extreme 15% for cognitive impairment, quasiautistic patterns, inattention/over activity, and disinhibited attachment (parent & teacher-report) | 58 < 6m, 59 6–24m, 48 > 24m | < 6m (combined with UK adoptees and non-PI Romanian adoptees), > 6m | logistic regression, Fisher’s exact test | yes | past work suggests step function at 6m | subnutrition did not relate to DSP independently or through an interaction with age at adoption; DSP was related to head circumference |
| Romania | Globally depriving | 6y, 11y, 15y | 144 | emotional (internalizing) and conduct (externalizing) problems | Rutter Scales (parent & teacher-report) | 58 < 6m, 59 6–24m, 48 > 24m | < 6m (combined with UK adoptees and non-PI Romanian adoptees), > 6m | ANOVA; path analysis | yes | past work suggests step function at 6m | no effect of gender; effect present at all ages, but interaction with age at assessment: more problems at 11 and 15 than at 6; mediated by DSP | |
| Ellis, Fisher, & Zaharie (2004) | Romania | — | 2–6y | 51 | anxiety/affective symptoms | ECI-4 (parent-report) | 1m–5.5y; M = 2.5y2 | continuous2 | multiple regression | yes | — | — |
| Elmund, Lindblad, Vinnerljung, & Hjern (2007) | Multiple | — | No info | 16,522 | out of home care due to child’s behavior | rates of out of home care in national database | 78% before age 2y, 15.5% age 2–3y, 6.5% age 4–6y | 0–1y, 2–3y, 4–6y | logistic regression | yes | — | — |
| Gunnar, van Dulmen, & the International Adoption Project Team (2007) | Multiple | — | 4–18y | 899 | behavior problems | CBCL (parent-report) | M = 29.5m, SD = 30.7m; 38% were ≥ 24m | < 12m, > 12m; < 24m, > 24m | multivariate logistic regression | yes | — | no interaction between age at adoption and PI status |
| Hawk & McCall (2011) | mostly Russia | Socially-emotionally depriving | 6–18y | 316 | behavior problems | CBCL (parent-report) | 3.3–57.7m; M = 16.0m, SD = 11.4 | < 18m, > 18m; < 6m, 6–11m, 12–17m, 18–23m, 24m+ | ANOVA, MANOVA | yes | step function at 18m | effects only present at older age at assessment (12–18y) |
| Merz & McCall (2010) | mostly Russia | Socially-emotionally depriving | 6–18y | 342 | behavior problems | CBCL (parent-report) | M = 16.4, SD = 11.75 | < 9m, 9–17m, ≥ 18m | chi square; hierarchical multiple regression | yes | step function at 18m | effects only present at older age at assessment (12–18y) |
| Verhulst, Althaus, & Versluis-Den Bieman (1990) | Multiple | — | 10–15y | 2148 | behavior problems | CBCL (parent-report) | 513 0–6m, 221 7–12m, 329 13–24m, 311 25–36m, 278 37–48m, 198 49–60m, 156 61–72m, 142 73–125m | < 6m, 7–12m, 13–24m, 25–36m, 37–48m, 49–60m, 61–72m, 73–125m | chi square; ANCOVAs | yes, for extreme scores (total sample), internalizing, and externalizing | sharp increase in problems after 24m | interaction with age at assessment for internalizing but not externalizing |
| Verhulst, Althaus, & Versluis-Den Bieman (1992) | Multiple | — | 9–17y | 3519 | behavior problems | CBCL (parent-report) | 0–10y; majority < 4y | < 6m, 7–12m, 13–24m, 25–36m, 37–48m, 49–60m, 61–72m, 73–125m | log-linear analyses | yes | — | effect is over and above that of age at assessment and gender; effect is better accounted for by abuse, neglect, and number of placements |
| Mixed Results | ||||||||||||
| Cederblad, Höök, Irhammer, & Merke (1999) | Multiple | — | 13–27y | 147 | behavior problems | CBCL (parent-report) | 54% < 7m; 30% > 1y | < 1y, > 1y at adoption | ANOVA | yes, for 1 subscale | — | — |
| Multiple | — | 13–27y | 147 | behavior problems | CBCL (parent-report) | 54% < 7m; 30% > 1y | < 6m or > 6m in institution or foster care | ANOVA | yes, for total score and 4 subscales | — | — | |
| Habersaat, Tessier, & Pierrehumbert (2011) | Multiple | — | 12–18y | 350 | behavior problems | CBCL (parent-report) | 83 < 6m, 52 6–12m, 43 12–24m, 172 > 24m | < 6m, 6–12m, 12–24m, > 24m | — | yes, for < 6m and > 24m; no for < 12m and > 12m | — | moderated by maternal solicitations style; not moderated by attachment |
| Kreppner, Rutter, Beckett, Castle, Colvert, Groothues, Hawkins, O’Connor, Stevens, & Sonuga-Barke (2007) | Romania | Globally depriving | 6y, 11y | 144 | impairments | most extreme 15% for cognitive impairment, quasiautistic patterns, inattention/over activity, disinhibited attachment, conduct problems, emotional problems, peerrelationship problems (parent & teacher-report) | 0–42m | continuous; < 6m, 6–24m, > 24m | chi square; binary logistic regression | yes, but only < 6m, > 6m comparison | step function at 6m | no interaction with weight at adoption (subnutrition) |
| Rutter, Kreppner, & O’Connor (2001) | Romania | Globally depriving | 6y | 165 | emotional (internalizing) and conduct (externalizing) problems | Rutter Scales (parent & teacher-report) | 58 < 6m, 59 6–24m, 48 > 24m | < 6m, 6–24m, 24–42m | chi square trends | no | — | — |
| Romania | Globally depriving | 6y | 165 | dysfunction | problems in attachment, attention, emotional, autism, cognition, peer or conduct domains (parent & teacher-report) | 58 < 6m, 59 6–24m, 48 > 24m | < 6m, 6–24m, 24–42m | — | yes | — | — | |
| Nielsen, Coleman, Guinn, & Robb (2004) | Ug and a | Various levels of deprivation | 1–3y | 33 | temperament | care giver interview | 3m–3y; M = 1.5y, SD = .67y2 | < 1y, 1–2y, 2–3y2 | MANCOVA | no | — | — |
| Ug and a | Various levels of deprivation | 1–3y | 33 | aggressiveness toward other children, objects, and adults | observation of free play | 3m–3y; M = 1.5y, SD = .67y2 | < 1y, 1–2y, 2–3y2 | MANCOVA | yes: only for interaction | — | interaction between duration of institutionalization and contact with relatives; longer institutionalization and contact with relatives was associated with aggressiveness toward adults | |
| Ug and a | Various levels of deprivation | 1–3y | 33 | negative affect, crying | observation of free play | 3m–3y; M = 1.5y, SD = .67y2 | < 1y, 1–2y, 2–3y2 | MANCOVA | yes: only for interaction | — | interaction between duration of institutionalization and contact with relatives; only multivariate effect reached significance | |
| Ug and a | Various levels of deprivation | 1–3y | 33 | positive affect | observation of free play | 3m–3y; M = 1.5y, SD = .67y2 | < 1y, 1–2y, 2–3y2 | MANCOVA | no | — | — | |
| Marcovitch, Cesaroni, Roberts, & Swanson (1995) | Romania | Globally depriving | 7m–11y | 130 | Behavior problems | questionnaire designed for current study (parent-report) | 100 < 2y, 30 > 2y | < 2y, > 2y | — | yes, for some domains (but no statistical test) | — | — |
| Tan, Marfo, & Dedrick (2007) | China3 | Socially-emotionally depriving | 1.5y–5y | SN = 63 Non-SN = 689 | behavior problems | CBCL 1.5–5y (parent-report) | SN: M = 25.85, SD = 10.82; non-SN: M = 12.71, SD = 4.87 | continuous | multiple regression | no | — | — |
| China3 | Socially-emotionally depriving | 6y–15.7y | SN = 61 Non-SN = 280 | behavior problems | CBCL (parent-report) | SN: M = 46.98, SD = 37.45; non-SN: M = 15.61, SD = 15.80 | continuous | multiple regression | yes, for Total problems (but small correlation) | — | — | |
| Tan, Marfo, & Dedrick (2010) | China3 | Socially-emotionally depriving | 1.5y–5y | 452 | behavior problems | CBCL (parent-report) | 3–50m; M = 13.1m, SD = 5.1m | continuous, natural log transformed | correlation, regression | yes for correlations at Time 2; only for Total (not internalizing, externalizing) at Time 1; for regression, only T2 Externalizing ant Total were significant (small magnitude) | — | — |
| Null Results | ||||||||||||
| Barcons-Castel, Fornieles-Deu, & Costas-Moragas (2011) | Multiple | — | 6–11y | 52 | behavior problems | various BASC subscales (parent & self-report) | 8–84m; M = 28.75m, SD = 21.42m | < 12m, 13–36m, > 37m | unifactorial ANOVA | no | — | — |
| Colvert, Rutter, Beckett, Castle, Groothues, Hawkins, Kreppner, O’Connor, Stevens, & Sonuga-Barke (2008a) | Romania | Globally depriving4 | 6y, 11y | 144 | emotional (internalizing) and conduct (externalizing) problems | Rutter Scales (parent & teacher-report) | < 43m | continuous; < 6m, > 6m | t-test, chi square | no | — | — |
| Kreppner, O’Connor, Rutter, & The ERA Study Team (2001) | Romania | Globally depriving | 4y, 6y | 137 | emotional (internalizing) and conduct (externalizing) problems | Rutter Scales (parent & teacher-report) | 0–42m | continuous; < 6m, 6–24m, > 24m | correlation | no | — | — |
| Cohen & Farnia (2011) | China3 | — | 1.5–4y | 70 | behavior problems | CBCL (parent-report) | 8–21m; M = 13m | continuous | regression | no: ns for internalizing, marginal for externalizing | — | — |
| Dalen & Rygold (2006) | China3 | — | 7–13y | 77 | problem behavior | Rutter scales (teacher-report) | 2–52m; M = 6.88m, SD = 9.28m | — | step-wise multiple regression | no | — | — |
| Elliott & McMahon (2011) | China3 | — | 3–8y | 59 | anxiety | PAS, SCAS-P (parent-report) | 6–54m; M = 14.71m, SD = 7.66m | continuous | correlation | no | — | — |
| Goldney, Donald, Sawyer, Kosky, & Priest (1996) | Indonesia | — | M = 15.2y, SD = 2.1y | 34 | behavior problems | CBCL (parent-report) | 1m–7.5y; M = 17m, SD = 1.7y | continuous | correlation | no | — | — |
| Indonesia | — | M = 15.2y, SD = 2.1y | 34 | behavior problems | YSR (self-report) | 1m–7.5y; M = 17m, SD = 1.7y | continuous | correlation | no | — | — | |
| Juffer & van Ijzen doorn (2005) | Multiple | — | all ages | 29 studies, 14,581 participants | behavior problems | various | 45% of studies 0–12m; 21% of studies 12–24m; 24% of studies > 24m; 10% of studies not reported | < 12m, > 12m; < 24m, > 24m | meta-analysis | no for Total, Internalizing, and Externalizing | — | — |
| Miller, Chan, Tirella, & Perrin (2009) | Eastern Europe | Globally depriving | 8–10y | 50 | behavior problems | BASC (parent & teacher-report) | 4–55m; 50% < 18m | continuous | correlation | no | — | — |
| Tan, Camras, Deng, Zhang, & Lu (2012) | China3 | Socially-emotionally depriving4 | 2–6y | 133 | behavior problems | CBCL (parent-report) | 8–31m; M = 12.8m | continuous | correlation | no for Total, Internalizing, and Externalizing | — | — |
| Tan & Marfo (2006) | China3 | Socially-emotionally depriving | 1.5y–11y | 695 | behavior problems | CBCL (parent-report) | preschool: M = 14.2m, SD = 8.3m; school-age: M = 18.5m, SD = 23.7m | continuous, log transformed | correlation, hierarchical regression | no for Total, Internalizing, and Externalizing | — | — |
| Tizard & Hodges (1978) | UK | Adequate | 8y | 25 | emotional (internalizing) and conduct (externalizing) problems | Rutter Scales (parent-report) | most between 2y and 4y; 5 between 4.5y and 7.5y | — | point biserial correlation | no | — | — |
| Tizard & Rees (1975) | UK | Adequate | 4.5y | 24 | emotional (internalizing) and conduct (externalizing) problems | Rutter Scales (parent-report) | Mean = 3.11y, SD = .69y | — | — | no | — | — |
| Age at admission: Distribution | Age at admission: Measurement | Significant effect of age at admission? | ||||||||||
| Lee, Seol, Sung, Miller, & the Minnesota International Adoption Project Team (2010) | South Korea | Adequate | 4–8y | 382 | behavior problems | CBCL (care giver-report) | M and SD for children with relinquishing parent, living parent, and deceased/unknown parent were 0.96y (0.64), 4.69y (1.50), 4.47y (1.46) | < 2y, > 2y | ANCOVA | yes | — | — |
| Wolkind (1974) | UK | — | 5y–12y | 92 | antisocial disorder, stealing, lying, destructiveness, fighting, disinhibition | caregiver interview | 49 < 2y, 43 > 2y | < 2y, > 2y | chi square, Fisher exact test | yes, only for disinhibition | — | effect for disinhibition only significant in older at assessment group (> 8y) |
| Vorria, Rutter, Pickles, Wolkind, & Hobsbaum (1998b) | Greece | Socially-emotionally depriving | 9–11y | 41 | emotional (internalizing) and conduct (externalizing) problems | Rutter Scales (parent & teacher-report) | nearly half admitted around 5–6y; 16 admitted before 3.5y | < 3y6m, 3y6m–5y6m, > 5y6m | — | no | — | — |
Note. Age at adoption findings are listed above age at admission findings. Within each portion of table, top section includes positive results, middle section includes mixed results, bottom section includes null results.
CBCL = Child Behavior Checklist, ECI-4 = Early Childhood Inventory-4, PAS = Preschool Anxiety Scale, SCAS-P = Spence Children’s Anxiety Scale-Parent Report, YSR = Youth Self Report, BASC = Behavioral Assessment System for Children, SN = special needs, DSP = Deprivation-specific psychological patterns
m = months, y = years, M = mean, Md = median, SD = standard deviation
This study used time in the orphanage instead of age at adoption; these two variables were correlated at r = .97
These studies assessed children who were still institutionalized, so duration of institutionalization was assessed rather than age at adoption
The vast majority of children adopted from China are female
These articles did not provide information about the conditions of these institutions, but other studies from the same authors on the same sample did describe the severity of the deprivation.
Severity of deprivation
The pattern of effects was highly related to the children’s region of origin. Of 13 studies (from 6 independent samples, including the ERA Study) of typically globally-depriving Eastern-European institutions, 7 studies (from 4 independent samples) demonstrated age at adoption effects in each analysis that was reported. Three additional Eastern-European studies (from 2 independent samples, including the ERA Study) detected effects in some analyses, but not others. Of 8 studies (from 5 independent samples) of children adopted from Asia, 2 studies from the same sample detected effects in only some analyses, and the remainder found only null results. A sample of children adopted from Western Europe (2 studies) failed to detect effects, and 1 study of children adopted from Uganda found mixed results. Because many studies don’t describe the quality of care provided in the institutions, the pattern of results with regard to severity of deprivation is consistent with expectations but less clearly defined.
Of two studies with a higher proportion of non-institutionalized children in their samples, one (Elmund, Lindblad, Vinnerljung, & Hjern, 2007) detected an age at adoption effect, and one (Marcovitch, Cesaroni, Roberts, & Swanson, 1995) detected an effect only for some domains of behavior problems.
Age at assessment
Several studies found age at adoption effects to be more predominant for children who were older at assessment (Hawk & McCall, 2011; Merz & McCall, 2010; Rutter et al., 2010; Verhulst et al., 1990). In one study, Romanian PI children had more externalizing and internalizing problems at age 11 and 15 than at age 6, though an age at adoption effect was detectable at each age at assessment (Rutter et al., 2010). When internalizing and externalizing behavior were examined separately, age at adoption interacted with age at assessment for internalizing, but not externalizing problems (Verhulst et al., 1990); in line with other studies, more problems were found at older ages at assessment.
Methodological issues
Whereas 11 of 15 studies that used the CBCL found significant effects for at least some analyses (8 out of 10 independent samples), only 1 of 7 studies (1 of 3 independent samples) that used the Rutter parent and teacher scales detected an effect. The CBCL and Rutter scales both tap into children’s externalizing and internalizing behavior problems and are highly correlated with each other (overall correlation r = 0.79; Fombonne, 1989 as cited in Elander & Rutter, 1995), so it is unclear why effects were more robust in studies using the CBCL. For studies that examined specific domains of behavior problems (e.g., subscales of the CBCL), there was no apparent pattern across studies as to which specific problems were more likely to show age at adoption effects. There was no apparent relationship between the likelihood of detecting an effect of age at adoption and the reporter (e.g., parent, teacher), number of subjects, the statistical analyses used, or the distribution or measurement of age at adoption.
Form of the relationship
Studies of children adopted from socially-emotionally depriving Russian institutions detected a step around 18 months of age at adoption (Hawk & McCall, 2011; Merz & McCall, 2010), but studies of children adopted from globally depriving Romanian institutions observe a similar step at 6 months of age at adoption (Kreppner et al., 2007; Rutter et al., 2010). One study of children adopted from a variety of countries detected a sharp increase in rates of problems for those adopted after 24 months of age at adoption (Verhulst et al., 1990). On the other hand, the age at adoption effect was linear for Romanian PI children for rocking, self-injury, and problems with chewing and swallowing (Beckett et al., 2002). In this same study, the age at adoption effect was a step function for unusual sensory interests (i.e., “an unusual interest in the feel and smell of persons or objects and unusual preoccupations with visual sensations”) assessed at 6 years of age; this behavior was significantly more common among children adopted after 6 months of age, consistent with other studies of children adopted from globally depriving Romanian institutions.
Associated variables
Age at adoption did not interact with whether a child had been reared in an institution or not (Gunnar, Van Dulmen, & The IAP Team, 2007), but because children in this study were from a variety of countries and levels of deprivation, the effects, if present, may be blurred. Age at adoption also did not appear to relate to subnutrition in Romanian PI children (Kreppner et al., 2007; Rutter et al., 2010). Brain growth, on the other hand, may relate to the effects of institutionalization on behavior problems; Rutter et al. (2010) found that DSP, which is elevated in children adopted after 6 months of age, is significantly related to head circumference. DSP mediated the age at adoption effect for internalizing, but not externalizing, behavior problems. One study reported that while age at adoption does relate to the level of maladjustment, the effect is better accounted for by abuse, neglect, and number of placements (Verhulst, Althaus, & Versluis-den Bieman, 1992).
In observations of institutionalized Ugandan children, longer institutionalization and visits from relatives were associated with aggressiveness toward adults; a similar multivariate interaction was detected for negative affect and crying, but no univariate effects were significant (Nielsen et al., 2004). The age at adoption effect was moderated by maternal interaction style, but not attachment, for adolescents adopted from multiple countries (Habersaat et al., 2011).
Age at admission effects
Children who enter institutions after 2 years of age may be less likely to demonstrate behavior problems than those who enter earlier; when age at admission cut-offs of 3.5 and 5.5 years were used, no age at admission effects were detected (Vorria et al., 1998b). One study found an age at admission effect for CBCL behavior problems (Lee, Seol, Sung, Miller, & Minnesota IAP Team, 2010), and one failed to detect age at admission effects for the Rutter scales (Vorria et al., 1998b); this pattern parallels the findings for age at adoption effects. An additional study assessed behavior problems with a parent interview and found that an age at admission effect (< 2 years, > 2 years) was detected for disinhibition, but not for antisocial disorder, stealing, lying, destructiveness, or fighting; effects were only evident in the older (8–12 years) age at assessment group (Wolkind, 1974), consistent with age at adoption findings.
Summary
Most studies found behavior problems to be more predominant for later-adoptees. Effects were more likely to be detected in studies of socially-emotionally or globally deprived children adopted from Eastern Europe, in subsamples assessed in adolescence rather than childhood, and when studies utilized the CBCL rather than the Rutter scales to assess behavior problems. The most commonly reported form of age at adoption effect appears to be a step function; the step seems to occur at 6 months of age at adoption for globally deprived Romanian children, and at 18 months of age at adoption for socially-emotionally deprived Russian children. But, for atypical behavior problems, there may be a linear relationship with age at adoption. Age at adoption effects do not appear to be accounted for by subnutrition or attachment, but head circumference, DSP, and maternal interaction style may play a role; contact with relatives may be a moderator for institutionalized children. While only limited evidence is available, children who enter institutional care in the first two years of life may have a higher risk for behavior problems than those who enter at a later age.
Attention Problems and Executive Function
Age at adoption effects for attention problems and executive function were assessed in 17 studies (10 independent samples); significant effects were found in 12 studies from 7 independent samples, and 4 studies (4 independent samples) found effects for only some analyses or subsamples (Table 8). In each case, later-adopted children had poorer outcomes.
Table 8.
Attention and Executive Function
| Study | Country of origin | Severity of deprivation | Age at assessment | N | Construct measured | Outcome measure | Age at adoption: Distribution | Age at adoption: Measurement | Statistical test | Significant effect of age at adoption? | Form of the relationship? | Relationship with other variables? |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Positive Results | ||||||||||||
| Audet & LeMare (2010) | Romania | Globally depriving1 | most 4.5y; some 5.5–9y | 46 | inattention/overactivity | CBCL subscale (parent-report) | 8–68m, Md = 18.5m | continuous | correlation | yes | — | no attenuation of effect overtime; role of parenting and attachment depend on duration of deprivation |
| Dalen (2001) | Colombia, Korea | — | 11–16y | 193 | hyperactive behavior | Rutter scales subscale (teacher-report) | 2–81m, M = 16.28m | 0–4m, 5–12m, 13–24m, 25m+ | — | yes | possible step function at 24m, but not tested | — |
| Gunnar, van Dulmen, & the International Adoption Project Team (2007) | Multiple | — | 4–18y | 899 | attention problems | CBCL subscale (parent-report) | M = 29.5m, SD = 30.7m; 38% were ≥ 24m | < 24m, > 24m | logistic regression | yes | — | — |
| Habersaat, Tessier, & Pierre humbert (2011) | Multiple | — | 12–18y | 350 | attention problems | CBCL subscale (parent-report) | 83 < 6m, 52 6–12m, 43 12–24m, 172 > 24m | < 6m, 6–12m, 12–24m, > 24m | — | yes, for < 6m and > 24m and < 12m and > 12m | — | — |
| Hawk & McCall (2011) | mostly Russia | Socially-emotionally depriving | 6–18y | 316 | attention problems | CBCL subscale (parent-report) | 3.3–57.7m; M = 16.0m, SD = 11.4 | < 18m, > 18m; < 6m, 6–11m, 12–17m, 18–23m, 24m+ | oneway ANOVA | yes | possible step function at 18m, but not tested | — |
| Merz & McCall (2010) | mostly Russia | Socially-emotionally depriving | 6–18y | 342 | attention problems | CBCL subscale (parent-report) | M = 16.4, SD = 11.75 | < 9m, 9–17m, ≥ 18m | chi square | yes | step function at 18m | age at adoption effect for older (12–18y) but not young er (6–11y) subsample |
| Kreppner, O’Connor, Rutter, & The ERA Study Team (2001) | Romania | Globally depriving | 4y, 6y | 137 | inattention/overactivity | Rutter scales subscale (parent & teacher-report) | 0–42m | continuous; < 6m, 6–24m, > 24m | correlation, Linear-by-Linear association test | yes | — | effects did not attenuate overtime; developmental level at admission was not a mediator; weight at admission mediated teacher-reported I/O but not parent-reported I/O; effects significant when sample restricted to normal range IQ |
| Rutter, Kreppner, & O’Connor (2001) | Romania | Globally depriving | 6y | 165 | inattention/overactivity | Rutter scales subscale (parent & teacher-report) | 58 < 6m, 59 6–24m, 48 24–42m | < 6m, 6–24m, > 24m | chi square trends | yes | — | — |
| Rutter, Sonuga-Barke, Beckett, Castle, Kreppner, Kumsta, Schlotz, Stevens, & Bell (2010) | Romania | Globally depriving | 6y, 11y, 15y | 144 | inattention/overactivity | Rutter scales subscale (parent & teacher-report) | 58 < 6m, 59 6–24m, 48 24–42m | < 6m (combined with UK adoptees and non-PI Romanian adoptees), > 6m | ANOVA, chi square | yes | — | — |
| Sonuga-Barke, Beckett, Kreppner, Castle, Colvert, Stevens, Hawkins, & Rutter (2008) | Romania | Globally depriving | 11y | 138 | inattention/overactivity | Rutter scales subscale (parent & teacher-report) | .5m–3y7m | < 6m, > 6m | ANOVA, mediational analyses, regression | yes | — | no interaction with subnutrition; overall sample: not mediated by head circumference; sub-nourished subgroup: possibly mediated by head circumference |
| Stevens, Sonuga-Barke, Kreppner, Beckett, Castle, Colvert, Groothues, Hawkins, & Rutter (2008) | Romania | Globally depriving | 11y | 144 | inattention/overactivity | Rutter scales subscale (parent & teacher-report) | 42 < 6m, 49 6–24m, 40 24–42m | < 6m, 6–24m, > 24m | ANOVA, chi square, post-hoc Tukey’s tests, t-test | yes | step function at 6m | — |
| Lindblad, Weitoft, & Hjern (2010) | Multiple | — | 6–21y | 16,134 | ADHD | ADHD medication rates | 70% < 2y | 0y, 1y, 2–3y, 4–7y | logistic regression | yes | — | Odds ratios similar for boys and girls |
| Mixed Results | ||||||||||||
| Barcons-Castel, Fornieles-Deu, & Costas-Moragas (2011) | Multiple | — | 6–11y | 52 | attentional problems | BASC subscale (parent & self-report) | 8–84m; M = 28.75m, SD = 21.42m | < 12m, 13–36m, > 37m | unifactorial ANOVA | yes | — | — |
| Multiple | — | 6–11y | 52 | hyper activity | BASC subscale (parent & self-report) | 8–84m; M = 28.75m, SD = 21.42m | < 12m, 13–36m, > 37m | unifactorial ANOVA | no | — | — | |
| Cederblad, Höök, Irhammer, & Merke (1999) | Multiple | — | 13–27y | 147 | attention problems | CBCL subscale (parent-report) | 54% < 7m, 30% > 1y | < 1y, > 1y at adoption and < 6m or > 6m in institution or foster care | two-way ANOVA | yes, for time in institution or foster care (not age at adoption) | — | — |
| Colvert, Rutter, Kreppner, Beckett, Castle, Groothues, Hawkins, Stevens, & Sonuga-Barke (2008b) | Romania | Globally depriving1 | 11y | 144 | executive function | Stroop task (experimenter tested) | 58 < 6m, 59 6–24m, 48 24–42m | continuous, < 6m, > 6m | correlation | yes for full sample, ns when restricted to > 6m | step function at 6m | not accounted for by IQ |
| Merz & McCall (2011) | mostly Russia | Socially-emotionally depriving | 5–18y | 288 | executive function | BRIEF (parent-report) | 5–60m, M = 15.7m, SD = 11.4m | < 9m, 9–17m, 18–26m, ≥ 27m | ANOVA, hierarchical multiple regression | yes | step function at 18m | interaction with age at assessment (highest risk: > 12y, adopted > 18m) |
| mostly Russia | Socially-emotionally depriving | 2–5y | 130 | executive function | BRIEF-P (parent-report) | 5–27m, M = 12.09m, SD = 4.12m | < 9m, 9–17m, 18–26m, ≥ 27m | correlation | no | — | — | |
| Null Results | ||||||||||||
| Bos, Fox, Zeanah, & Nelson (2009) | Romania | Globally depriving1 | 8y | < 93 | executive function | CANTAB (experimenter tested) | 9–33m, M = 23.6m2 | continuous; dichotomized at 2m intervals from 18m to 28m (e.g., < 18m, > 18m; < 20m, > 20m)2 | independent samples t-tests; correlation | no | — | — |
Note. Top section includes positive results, middle section includes mixed results, bottom section includes null results
CBCL = Child Behavior Checklist, BRIEF = Behavior Rating Inventory of Executive Function, BRIEF-P = Behavior Rating Inventory of Executive Function – Preschool Version, BASC = Behavioral Assessment System for Children CANTAB = Cambridge Neuropsychological Test and Automated Battery,
I/O = inattention/overactivity, m = months, y = years, M = mean, Md = median, SD = standard deviation
This article did not provide information about the conditions of these institutions, but other studies from the same authors on the same sample did describe the severity of the deprivation.
In this study, children moved from an institution to foster care (not an adoptive home) at this age
Severity of deprivation
Severity of deprivation did not appear to differentiate between studies that detected effects of age at adoption and those that did not.
Age at assessment
One sample found age at adoption effects in older (12–18 year old, and 5–18 year old), but not younger (6–11 year old, and 2–5 year old) subsamples (Merz & McCall, 2010, 2011). In two additional samples, the age at adoption effect did not attenuate over time (Audet & LeMare, 2010; Kreppner, O’Connor, Rutter, & The ERA Study Team, 2001).
Methodological issues
The only study that failed to detect an effect of age at adoption on executive function (Bos et al., 2009) assessed 8-year-old Romanian PI children with the CANTAB, which is a touch-screen automated neuropsychological battery that taps into visual memory, visual attention, and planning/working memory; it could be that this measure did not assess the specific deficits exhibited by this population. In this study, children moved from institutions to foster care (not adoptive homes), and children departed the institutions at 9 to 33 months of age (mean = 23.6 months). The most probable explanation for the null findings is that this sample was adopted too late (i.e., after 6 months of age) to detect age at adoption effects.
An additional study (Barcons-Castel, Fornieles-Deu, & Costas-Moragas, 2011) found that while later-adopted children had poorer scores on the attentional problems subscale of the BASC than earlier-adopted children, there was no difference on the hyperactivity subscale. Note, however, that a study that used the Rutter scales did find an age at adoption effect for hyperactivity (Dalen, 2001). There was no apparent relationship between the likelihood of detecting an age at adoption effect and the reporter (e.g., parent, teacher), number of subjects, the statistical analyses, or the measurement of age at adoption.
Form of the relationship
For children adopted from globally depriving Romanian institutions, a step function occurs around 6 months of age at adoption (Colvert et al., 2008b; Stevens et al., 2008). Children adopted from socially-emotionally depriving Russian institutions appear to have a step at 18 months of age at adoption (Hawk & McCall, 2011; Merz & McCall, 2010, 2011). While the level of deprivation was not reported, a study of children adopted from Colombia and Korea reported data consistent with a step function around 24 months of age at adoption, although a step function was not explicitly tested (Dalen, 2001).
Associated variables
Age at adoption effects were independent of IQ (Colvert et al., 2008b; Kreppner et al., 2001) and developmental level at adoption (Kreppner et al., 2001). Weight at adoption appeared to mediate the age at adoption effect for teacher-reported, and not parent-reported, inattention/overactivity (I/O; Kreppner et al., 2001). Age at adoption did not interact with subnutrition, but there was evidence of head circumference as a mediator of the age at adoption effect on I/O for subnourished (but not non-subnourished) children (Sonuga-Barke et al., 2008). Parenting variables like warmth, stimulation in the adoptive home, and attachment, are related to I/O after taking into account age at adoption and earlier I/O (Audet & LeMare, 2010). Further, for earlier-adopted children, authoritarian parenting positively predicted I/O, but for later-adopted children, authoritarian parenting negatively predicted I/O. One study found that odds ratios for the age at adoption effect on rates of ADHD medication use were similar for boys and girls (Lindblad, Weitoft & Hjern, 2010).
Summary
The vast majority of studies detected an age at adoption effect for attention and executive function, with later-adoptees faring more poorly. Age at adoption effects may be more likely to be detected in adolescence, and less likely in analyses of Romanian adoptees that include only children adopted after 6 months of age. The age at adoption effect appears to be a step function, with a step occurring at 6 months of age at adoption for globally deprived Romanian adoptees, 18 months of age at adoption for socially-emotionally deprived Russian adoptees, and 24 months of age at adoption for adoptees from Colombia and Korea. The age at adoption effect was not accounted for by developmental level, IQ, or subnutrition, but it may be partially mediated by head circumference in a subnourished subgroup.
Discussion
Age at adoption effects were detected in most studies across multiple outcome domains, but roughly a third of studies did not detect an effect; thus, a number of factors can be identified that relate to the likelihood of detecting age at adoption effects. Studies of globally-deprived children are most likely to find age at adoption effects, and in each study where age at adoption effects are detected, later-adoptees have more difficulties. Several independent samples (ERA Study, St. Petersburg USA Study Team, Verhulst’s studies) found that the age at adoption effect is more likely to be detected in adolescent than preschool and middle-childhood samples, so the effects do not seem to fade with increased age or time in the adoptive home. Age at adoption effects for behavior problems were more evident on the CBCL than on the Rutter scales, parent interviews were more likely than observational or self-report measures to detect age at adoption effects for attachment and RAD/DSB, and self-report measures were most likely to detect age at adoption effects for peer relationships.
With few exceptions, each study that examined the form of the age at adoption effect detected a step-like relationship between age at adoption and outcomes. Children adopted before a certain cut-off age are generally similar to parent-reared children in the problems they experience, but children adopted after that point have elevated rates of problems; continued institutionalization beyond this point does not seem to contribute to further increases in problems. Note, however, that research on the form of the age at adoption effect after about 42 months of age at adoption is rare, so conclusions only apply to children adopted before that age. Likewise, research is limited with regard to the age that children were admitted to the institution. While age at adoption and time in institution are typically highly correlated (see Table 1), most studies do not report whether children were institutionalized for the entirety of the period of time that may encompass a sensitive period.
The specific age at which the step occurred differed across studies in accord with the severity of deprivation and/or the country of origin. Studies of children adopted from globally depriving Romanian institutions tended to find a step function at 6 months of age at adoption, and studies of children from socially-emotionally depriving Russian institutions detected a step function at about 18 months of age at adoption. There were two main exceptions to this rule. First, the age at adoption effect for attachment and parent-child relationships may be linear, and some evidence suggests that effects are due to the contrast between children who leave the institution before versus after 24 months of age, even for globally-deprived children. Further, atypical behavior problems (e.g., rocking, self-injury, and problems with chewing and swallowing) may be linearly related to age at adoption. The etiology of these problems may differ from that of those problems that did have a clear step-like relationship to age at adoption.
Several studies examined the role of other variables in mediating or moderating the effect of age at adoption on outcomes. Cognitive ability, IQ, and developmental level at adoption failed to consistently account for age at adoption effects. For both disinhibited social behavior and attention and executive function outcomes, subnutrition did not interact with age at adoption, but there was some evidence that within a subnourished subsample, head circumference may mediate the age at adoption effect. Additionally, the effect of age at adoption on disinhibited social behavior was mediated by inhibitory control. This, in conjunction with the finding that age at adoption effects were most robust for attention and executive function compared to other outcomes, suggests that attention and executive function may represent a sort of “core deficit” that contributes to PI children’s deficits in other domains.
In order to have a complete understanding of the influence of the timing of early experience, it is essential to know when such experience started in addition to when it ended. Unfortunately, information on age at admission to institutional care is seldom available or studied. The data that are available suggests that children who enter before 2 years of age have more problems than children who enter after this age. Deprivation within the first two years of life seems to be particularly important to children’s development, so it is likely that timing of deprivation rather than simple duration of exposure, contributes to problems later on. This is consistent with primate research that shows that earlier social deprivation has much more devastating effects than later deprivation (Harlow & Novak, 1973; Suomi, 2003). Thus, a child who experienced deprivation from birth to 2 years of age might be expected to have more difficulties later on than a child who experienced similar levels of deprivation from 2 to 4 years of age.
Reconsidering Hypotheses of How Early Deprivation Affects Development
Cumulative Effects
According to the cumulative effects hypothesis, individuals are the product of their environment and are highly malleable, so the effects of institutionalization should increase linearly with duration of institutionalization and fade with increased age or time in the adoptive home. But, institutionalization in the first two years of life seems to be particularly detrimental, and increased time in the institution beyond this point does not seem to further contribute to risk of additional problems. The form of the age at adoption effect appears to be a step-function, not linear, for most outcomes. While some studies have detected age at adoption effects soon after adoption (e.g., Tan’s studies), most studies find that problems are more evident in adolescence.
The prediction of this hypothesis that quality of care should relate to outcomes seems to be partially met by the available data; it takes less time in a globally depriving institution than in a socially-emotionally depriving or adequate institution to produce noticeable effects on the risk of problems in adopted children. But, evidence is mixed as to whether quality of care in the adoptive home contributes significantly to outcomes.
The cumulative effects hypothesis may hold true for some children, or for some outcomes. Massive catch-up growth occurs in the first few years after adoption (van IJzendoorn et al., 2007; van IJzendoorn & Juffer, 2006), attesting to the benefits provided by moving to a supportive environment. However, benefits do not continue to grow with more time in the adoptive home. This is not to say that the adoptive home environment doesn’t matter after the initial adjustment period (it does—see Audet & LeMare, 2010), but rather that the duration of time in the adoptive home fails to account for outcomes. It remains possible that for some children, problems were evident soon after adoption, but faded with increased time in a supportive home environment, consistent with the predictions of the cumulative effects hypothesis.
As predicted, a linear age at adoption effect was found for some atypical behavior problems, attachment disorder behavior, and possibly for attachment organization. For atypical behavior problems, there tended to be fewer problems at older ages at assessment, consistent with the hypothesized decline in problems with more time in a supportive environment. While the cumulative effects hypothesis was not supported for most outcome variables, this hypothesis may contribute to our understanding of the effects of institutionalization on attachment and atypical behavior problems.
Developmental Programming
The two developmental programming hypotheses (experience-expectant and experience-adaptive) predict that more time in a supportive adoptive home or an older age at assessment should not be linked to fewer problems because the effects of institutionalization are thought to be permanent, and they propose that there may be sensitive periods of exposure, so a step function would be likely. These predictions are largely met by the data. Early institutional experience is related to elevated rates of problems, and when deprivation is severe, problems often result even when exposure is limited to the first 6–12 months of life. Moreover, the effects of institutionalization are detected years after adoption, and age at adoption effects are somewhat more likely to be detected in adolescence than in childhood; while the effects of institutionalization appear to be permanent, there may be a sleeper effect. A step function is evident in most studies that described the form of the age at adoption function, but the specific age at which the step occurs varies according to the severity of institutional deprivation.
Experience-expectant developmental programming
The experience-expectant developmental programming hypothesis posits that there are specific “necessary experiences” that a child must have in order for development to proceed properly, and permanent deficits will necessarily result if these experiences are not had. The available evidence does not fit these expectations. For studies of socially-emotionally deprived (Hawk & McCall, 2011) and globally deprived (Colvert et al., 2008a; Kreppner et al., 2007) children, only up to half of later-adopted children have problems. It is possible that only some children lacked “necessary experience” in an institution, but it is more likely that some children are resilient in the face of deprivation.
Experience-adaptive developmental programming
The experience-adaptive developmental programming hypothesis suggests that during early sensitive periods, children adapt to the environment they reside in, such that a drastic change in environment (e.g., by adoption) could render a child maladapted to his new environment. For most outcomes, this hypothesis is not supported; difficult parent-child relationships, poor theory of mind or quasi-autistic patterns, peer problems, social problems, behavior problems, and executive function problems do not seem likely to be helpful to a child residing in an institution, so these behaviors are not expected to be the consequence of appropriate adaptation to the institutional environment. Disinhibited social behavior might represent a behavior that is adaptive in the institution because it can help a child gain attention from caregivers, but if this behavior is, in fact, rooted in difficulties with inhibitory control (Bruce et al., 2009), it is less likely to be an adaptive response. The experience-adaptive hypothesis also suggests that variations in the specific level and type of deprivation that a child experiences could relate to outcomes. This prediction appears to be met because the levels of problems tend to vary on a continuum according to the level of deprivation children experienced in the institution.
Summary
None of the above hypotheses fully account for the effects of early institutionalization on development. The cumulative effects hypothesis appears to fit for some atypical behavior problems and for some attachment outcomes, and many aspects of developmental programming are supported by the available data, but the specific hypotheses (experience-expectant and experience-adaptive) fall short. Consistent with the developmental programming hypotheses, institutionalization effects are long-lasting, and the ages of exposure are important in predicting risk of problems. But, several questions remain.
Issues to Consider
Sensitive periods
The current review suggests that sensitive periods4 may play a role in social and behavioral development because when studies examined the form of the age at adoption function, they tended to find a step function. Traditional notions of sensitive periods specify that there is a distinct period of elevated vulnerability to environmental influences, and this period is universal within a species. For institutional deprivation, however, the boundaries of the period of vulnerability vary with the severity of institutional deprivation to which a child is exposed. While most accounts of sensitive periods describe the temporal boundaries to be inflexible, more recent accounts suggest that sensory deprivation can prolong the opening of a sensitive period, or compensatory processes can mask the closing of a sensitive period (Knudsen, 2004; Zeanah, Gunnar, McCall, Kreppner, & Fox, 2011). It is possible that such processes contribute to the variability in the temporal boundaries of sensitive periods that is suggested by this review.
The existence of a sensitive period in itself does not imply a particular causal mechanism, but the step function may imply some kind of intraorganismic change (Kreppner et al., 2007). There is insufficient evidence to be certain about the changes that mark the end of a potential sensitive period, but epigenetic effects are likely to be partially responsible. Early elevated stress levels are linked to epigenetic changes (Shonkoff, Boyce, & McEwen, 2009) that, in animal studies, are associated with prolonged stress responses, changes in brain architecture and chemistry, and behaviors resembling depression and anxiety (National Scientific Council on the Developing Child, 2010). If a similar model exists for humans, a certain amount of deprivation might be required before epigenetic changes take place; this threshold may be reached at an earlier age (e.g., 6 months) in globally depriving institutions, and a later age (e.g., 18 months) in socially-emotionally depriving institutions. While the cumulative effects hypothesis alone did not adequately account for the data, cumulative exposure in a broad sensitive period of the first two years of life might contribute to the lasting effects of early experience.
Resilience and individual differences
Our current hypotheses propose that anyone exposed to deprivation, especially severe deprivation, should exhibit its negative effects. In fact, only a portion of children, even from the most depriving institutions, demonstrate negative outcomes. PI children are remarkably resilient despite the severity of their deprivation.
Several factors might account for the resiliency of children adopted from depriving institutions. Individual differences in quality of care relate to individual differences in outcomes (Smyke et al., 2007), so children who have better outcomes may be those who were “favorites” in the institution. But, particularly in globally depriving institutions, this is unlikely to account for the positive outcomes of many children. Animal models suggest that positive experiences prior to deprivation might buffer children from poor outcomes, but most children enter institutional care very early in life, so this again is unlikely to account for the positive outcomes of many children. It is possible that better prenatal circumstances might contribute to resilience in some children, but most studies find that broad factors like birthweight or prematurity do not significantly account for differences in outcomes among this population (Kreppner et al., 2007; Merz & McCall, 2010, 2011; Sonuga-Barke et al., 2008). It remains possible that other prenatal factors (e.g., nutrition, stress, alcohol/substance use, etc.) may play a role, but more research is needed.
A final possibility is that genetic factors contribute to resilience (Caspi et al., 2003). Genotype by duration of institutionalization interactions may play a role for some domains of problems (e.g., attention or emotional problems; van IJzendoorn et al., 2011). Genes may also play a role in determining the degree to which a child is sensitive to the environment (National Scientific Council on the Developing Child, 2005), such that children who are resilient to the effects of institutionalization may be those who are genetically least vulnerable to environmental effects (Rutter, 2003; Rutter, Moffit, & Caspi, 2006).
Sleeper effects
Age at adoption effects, and thus effects of institutionalization, appear to be most evident among adolescent samples, suggesting that institutionalization-related problems are not simply the persistence of learned behaviors; it is more likely that problems are reflective of underlying deficits in certain basic skills that are not evident until relatively later in development (Zeanah et al., 2011). Apparent sleeper effects have previously been detected for effects of intervention programs (Reynolds & Robertson, 2003; Olds et al., 1997) and variously attributed to increases in statistical power over time (Mersky, Topitzes, & Reynolds, 2011), or the likelihood that some intervention effects simply take time to emerge (Reynolds & Robertson, 2003). Alternatively, early neural changes may contribute to sleeper effects. For example, early visual deprivation might be associated with a failure to establish optimal neural architecture for future visual development (Maurer, Mondloch, & Lewis, 2007). Thus, a lack of specific types of early experience might lead to neural changes or prevent neural architecture from developing in expected ways, and these changes may affect the development of certain skills that do not emerge until later on in development. Alternatively, it could be that problems emerge for this population in adolescence due to the elevated social and behavioral demands and expectations at this point in development. Residence in a supportive adoptive home environment may provide adoptees with the necessary support and guidance to prevent many of these issues from becoming problematic in their early years. When adults begin to pull back their support, as is typical for any child entering adolescence, more problems may emerge.
What is it about the institution?
It is widely accepted that institutionalization is associated with risk for poor developmental outcomes, but less is known regarding what, specifically, it is about the institution that produces such negative effects on development. The behavioral environment of institutions (specifically the lack of caregiver-child relationships), is typically hypothesized to be causally related to outcomes, but other potential explanations must be addressed as well. First, it could be that the less healthy children are adopted at later ages. But, age at adoption effects are evident for 1990s Romanian adoptees, and these children were adopted en masse after the fall of the Ceausescu regime; thus, in studies of Romanian adoptees, age at adoption is largely a function of a child’s age at the fall of the Ceausescu regime rather than any characteristics of the child (see p. 6, Rutter et al., 2010). Furthermore, most studies have not detected a significant relationship between birthweight or prematurity and outcome variables (Kreppner et al., 2007; Merz & McCall, 2010, 2011; Sonuga-Barke et al., 2008; but see also Bos et al., 2009). Prenatal factors might account for some between-country differences; because many Chinese children are abandoned due to the One Child Policy, they may have had better prenatal circumstances than children who were relinquished due to poverty or drug abuse (Cohen & Farnia, 2011), and this could partially account for the relatively more favorable outcomes for Chinese PI children. There also may be between-country differences in the age at which children are admitted to institutions and the quality of care children receive prior to admission, but little is currently known about these factors.
While globally depriving institutions such as those in Romania were nutritionally and medically deficient in addition to being socially-emotionally deficient, it is not likely that these factors are primary in producing effects of institutionalization. Subnutrition was not found to account for age at adoption effects, and lasting effects of institutionalization are found even for children who resided in institutions with adequate nutrition and medical care. It is possible, however, that children who had poorer nutrition (e.g., those in globally depriving Romanian institutions) were more vulnerable to the negative effects of institutionalization (Rutter, 2005b).
Another possibility is that the institutional experience is quite stressful for infants and young children, and the biological and psychological sequelae of early stressful experiences contribute to children’s poor developmental outcomes. Early sensitive and responsive caregiving, which is lacking in institutions, maintains a hyporesponsive neuroendocrine stress system (Gunnar et al., 2006). Early stressful experiences, particularly in the absence of a secure attachment relationship, are linked to disturbances in diurnal cortisol patterns (Kertes, Gunnar, Madsen, & Long, 2008) and problems with executive function and emotion regulation (which are also deficient in PI populations), and can produce effects that last through childhood and adulthood (National Scientific Council on the Developing Child, 2008, 2010, 2011). This is likely to account for some of the difficulties encountered by children reared in institutions, but it is unable to account for problems like disinhibited social behavior, which are relatively common in PI populations, but rare in other populations that experience early stress (Rutter, 2005b).
Given the evidence, it is likely that the distinct lack of early caregiver-child relationships in institutions is a primary contributor to the lasting problems that PI children experience. Even in institutions where medical, nutrition, and safety needs are met, there tend to be many and changing caregivers who provide care in a perfunctory way. Children typically do not see the same caregivers from day to day, and caregiver-child relationships rarely develop. Caregiver-child relationships are fundamental to early development; children seek comfort and emotional support from caregivers, and also learn basic skills like an understanding of contingencies, social cues, and social agency from stimulation and interactions with a sensitive and responsive caregiver (Sroufe & Waters, 1977). Deficient early caregiver-child relationships are likely to contribute to the lack of social inhibition and awareness that characterizes disinhibited social behavior. Relationship deprivation also contributes to elevated stress levels in infants and limits the amount and quality of their environmental and relationship stimulation. Furthermore, individual differences in quality of care are related to individual differences in outcomes (Smyke et al., 2007), and when early caregiver-child relationships are improved through an institutional intervention (The St. Petersburg-USA Orphanage Research Team, 2008) or through transfer to foster care (Smyke et al., 2010), developmental outcomes improve remarkably.
Limitations
Several limitations are apparent with respect to the data available to address the major questions of this review. First, most children enter institutional care in the first weeks or months of life, so it is impossible to separate ages of exposure from duration of institutional deprivation. While there are a few studies of age at admission that suggest that children in the first two years of life are particularly vulnerable to the effects of institutionalization, most research is unable to make such a distinction. While the current literature suggests a possible sensitive period, it is often not known whether children resided in an institution for the entirety of that period or experienced other placements. Further research is needed to tease apart ages and duration of exposure and their relation to outcomes. Similarly, most research focuses exclusively on the timing and effects of earlier negative experiences but not later positive experiences; while development is a product of both negative and positive experiences, this latter component of experience is rarely thoroughly addressed. Little to no data tends to be available regarding children’s pre-institutional experience, including other placements (e.g., hospitals, foster care, other institutions) and prenatal experience, so these factors cannot be ruled out as contributors to the effects of institutionalization and age at adoption. But, studies of domestic adoption show that even when prenatal experience is quite poor (e.g., in the case of prenatal drug exposure), postnatal experience plays a primary role in children’s outcomes (Rutter, 2005a). Additionally, information is often not available about specific characteristics of various institutions, much less about a specific child’s experience in the institution, so it is impossible to determine whether individual differences in outcomes relate to individuals’ different experiences in institutions.
Furthermore, a large number of studies were included in this review, but a smaller number of independent samples are available. Conclusions about the relationship between age at adoption and other variables and the form of the age at adoption effect have largely not been replicated in independent samples and are based on a limited number of studies. When the form of the age at adoption function was examined, it was not always fully examined; studies that reported a linear relationship did not always test whether a step function (or another function) fit better, and vice versa.
Conclusions
Despite these limitations, the message from this set of studies is clear. Children adopted from institutions have higher rates of problems when they are adopted at later ages, and for most outcomes, institutionalization beyond a certain age is associated with a step-like increase in risk for lasting social and behavioral problems. The age at which a step occurs seems to vary with the severity of the institutional deprivation, with a step apparent at 6 months of age at adoption for globally deprived Romanian adoptees and 18 months of age at adoption for socially-emotionally deprived Russian adoptees. A later age at admission into an institution is associated with lower rates of problems, suggesting the importance of experience in the first two years of life (rather than simple duration of experience) in predicting later outcomes. The step function suggests that sensitive periods may play a role, and evidence suggests that many (even later-adopted) children are remarkably resilient, but these findings are not well explained by our current hypotheses. More research is needed to separate the effects of ages and duration of exposure to the institutional environment.
The current state of the literature does not support firm conclusions about causes for the patterns detected in the data, but some speculative explanations can be offered. In institutions, deficient caregiver-child relationships are likely to contribute to children’s limited environmental and relationship stimulation, which likely relates to the disinhibited social behavior that is often seen in PI populations. Relationship deprivation is also expected to contribute to a diminished buffering of institutionalized children’s stress response, which means children would be exposed to elevated cortisol levels at a time when they are particularly vulnerable. Early stress experience is known to relate to later attention, executive function, and emotion regulation problems, which are common problems in PI populations and may contribute to other behavior problems.
Furthermore, genetic factors may relate to the step function that is detected and to the resilience of some children. Because of their genetic make-up, some children may be less vulnerable to the effects of their environment, rendering them less likely to show negative effects of institutionalization in the long-term. For those who are more vulnerable to environmental influences, epigenetic changes may be triggered once a certain threshold of deprivation has been reached. Globally deprived Romanian children may experience epigenetic changes as early as 6 months of age, whereas similar epigenetic changes may not occur until closer to 18 months of age when the deprivation is solely social-emotional in nature.
Future Directions
Several factors should be addressed in order for future studies to more completely address the question of the impact of early experience on later development. When possible, studies should examine the potential influence of prenatal and pre-institutional circumstances, other placements, and individual differences in quality of care received on the later development of PI children. When there is sufficient variability in age at admission to the institution, age at admission effects should be examined to contribute to our understanding of the importance of specific ages of exposure to deprivation, and efforts should be made to distinguish the effects of duration of exposure from specific ages of exposure to institutionalization. Future studies should provide more thorough and complete assessments of the effects of age at adoption; for example, studies could test both linear and step functions (using multiple possible “cut-off” ages at adoption) for multiple outcome measures, and include tests to determine the best fitting form of the age at adoption function.
More research is needed to explore potential mechanisms to account for sensitive periods and resiliency in this population of children. Specifically, future studies should explore the possible influence of stress hormones and genetics and the role they may play in producing a step function for age at adoption and individual differences between children. Our understanding of the processes of development would benefit from studies aimed to measure children’s development during the time they reside in an institution; ideally, such studies would identify when specific aspects of development and biology change, and connect these data to outcomes in childhood and adolescence.
Acknowledgments
This research was supported by NICHD grants HD39017 and HD050212 to Robert B. McCall and Christina J. Groark. The content is solely the responsibility of the author and does not necessarily represent the official views of the NICHD or NIH. The author is grateful to Robert B. McCall, Celia Brownell, Susan B. Campbell, Daniel S. Shaw, and Junlei Li for their helpful comments on earlier drafts of this manuscript.
Abbreviations
- PI
post-institutionalized
- DSP
deprivation-specific problems
- I/O
inattention/overactivity
Footnotes
These meta-analyses utilized an internationally adopted sample that was not exclusively institution-reared, but most studies included a majority of institution-reared children.
An additional meta-analysis (van den Dries et al., 2009) did not report whether participants were institution-reared, but will still be included in this review because the majority of component studies focused on institution-reared children.
Over half of the Chinese sample experienced an average of 8.4 months of foster care, so it is possible that this Chinese sample was less deprived than otherwise might be expected.
A strict understanding of sensitive periods suggests that sensitive periods lie at the level of circuits, not complex behaviors like those described here (Zeanah, Gunnar, McCall, Kreppner, & Fox, 2011). While this discussion considers only complex behaviors, it is understood that multiple circuit-level sensitive periods underlie these behaviors.
References
- Ames . The development of Romanian orphanage children adopted to Canada: Final Report. Simon Fraser University; Vancouver, Canada: 1997. [Google Scholar]
- Anderson CO, Mason WA. Early experience and complexity of social organization in groups of young rhesus monkeys. Journal of Comparative and Physiological Psychology. 1974;87(4):681–690. [Google Scholar]
- Audet K, Le Mare L. Mitigating effects of the adoptive caregiving environment on inattention/overactivity in children adopted from Romanian orphanages. International Journal of Behavioral Development. 2010;35(2):107–115. doi: 10.1177/0165025410373313. [DOI] [Google Scholar]
- Barcons N, Abrines N, Brun C, Sartini C, Fumado V, Marre D. Social relationships in children from intercountry adoption. Children and Youth Services Review. 2012;34:955–961. doi: 10.1016/j.childyouth.2012.01.028. [DOI] [Google Scholar]
- Barcons-Castel N, Fornieles-Deu A, Costas-Moragas C. International adoption: Assessment of adaptive and maladaptive behavior of adopted minors in Spain. The Spanish Journal of Psychology. 2011;14(1):123–132. doi: 10.5209/rev_SJOP.2011v14.n1.10. [DOI] [PubMed] [Google Scholar]
- Beckett C, Bredenkamp D, Castle J, Groothues C, O’Connor TG, Rutter M, The English and Romanian Adoptees Study Team Behavior patterns associated with institutional deprivation: a study of children adopted from Romania. Journal of Developmental and Behavioral Pediatrics. 2002;23(5):297–303. doi: 10.1097/00004703-200210000-00001. [DOI] [PubMed] [Google Scholar]
- Bos KJ, Fox N, Zeanah CH, Nelson CA. Effects of early psychosocial deprivation on the development of memory and executive function. Frontiers in Behavioral Neuroscience. 2009;3:1–16. doi: 10.3389/neuro.08.016.2009. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bowlby J. Maternal care and mental health. Geneva: World Health Organization; 1951. (WHO Monograph No. 2) [PMC free article] [PubMed] [Google Scholar]
- Bronfenbrenner U. The ecology of human development: Experiments by nature and design. Cambridge: Harvard University Press; 1979. [Google Scholar]
- Bruce J, Tarullo AR, Gunnar MR. Disinhibited social behavior among internationally adopted children. Development and Psychopathology. 2009;21(1):157–171. doi: 10.1017/S0954579409000108. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Camras LA, Perlman SB, Wismer Fries AB, Pollak SD. Post-institutionalized Chinese and Eastern European children: Heterogeneity in the development of emotion understanding. International Journal of Behavioral Development. 2006;30(3):193–199. doi: 10.1177/0165025406063608. [DOI] [Google Scholar]
- Caspi A, Sugden K, Moffitt TE, Taylor A, Craig IW, Harrington H, Poulton R. Influence of life stress on depression: Moderation by a polymorphism in the 5-HTT Gene. Science. 2003;301(5631):386–389. doi: 10.1126/science.1083968. [DOI] [PubMed] [Google Scholar]
- Castle J, Groothues C, Bredenkamp D, Beckett C, O’Connor TG, Rutter M. Effects of qualities of early institutional care on cognitive attainment. American Journal of Orthopsychiatry. 1999;69(4):424–437. doi: 10.1037/h0080391. [DOI] [PubMed] [Google Scholar]
- Cederblad M, Höök B, Irhammar M, Mercke A. Mental health in international adoptees as teenagers and young adults. An epidemiological study. Journal of Child Psychology and Psychiatry. 1999;40(8):1239–1248. [PubMed] [Google Scholar]
- Chisholm K. A three year follow-up of attachment and indiscriminate friendliness in children adopted from Romanian orphanages. Child Development. 1998;69(4):1092–1106. [PubMed] [Google Scholar]
- Chisholm K, Carter MC, Ames EW, Morison SJ. Attachment security and indiscriminately friendly behavior in children adopted from Romanian orphanages. Development and Psychopathology. 1995;7:283–294. [Google Scholar]
- Clarke AM, Clarke ADB. Early experience and the life path. London: Jessica Kingsley; 2000. [Google Scholar]
- Cohen NJ, Farnia F. Social-emotional adjustment and attachment in children adopted from China: Processes and predictors of change. International Journal of Behavioral Development. 2011;35(1):67–77. doi: 10.1177/0165025410371602. [DOI] [Google Scholar]
- Colvert E, Rutter M, Beckett C, Castle J, Groothues C, Hawkins A, Sonuga-Barke EJS. Emotional difficulties in early adolescence following severe early deprivation: Findings from the English and Romanian adoptees study. Development and Psychopathology. 2008a;20(2):547–567. doi: 10.1017/S0954579408000278. [DOI] [PubMed] [Google Scholar]
- Colvert E, Rutter M, Kreppner J, Beckett C, Castle J, Groothues C, Sonuga-Barke EJS. Do theory of mind and executive function deficits underlie the adverse outcomes associated with profound early deprivation?: Findings from the English and Romanian Adoptees study. Journal of Abnormal Child Psychology. 2008b;36(7):1057–1068. doi: 10.1007/s10802-008-9232-x. [DOI] [PubMed] [Google Scholar]
- Croft C, O’Connor TG, Keaveney L, Groothues C, Rutter M, The English and Romanian Adoption Study Team Longitudinal change in parenting associated with developmental delay and catch-up. Journal of Child Psychology and Psychiatry. 2001;42(5):649–659. [PubMed] [Google Scholar]
- Dalen M. School performances among internationally adopted children in Norway. Adoption Quarterly. 2001;5(2):39–57. [Google Scholar]
- Dalen M, Rygold A. Educational achievement in adopted children from China. Adoption Quarterly. 2006;9(4):45–58. doi: 10.1300/J145v09n04_03. [DOI] [Google Scholar]
- Decker S, Omori M. Age at adoption: Long-term measures of success in adulthood. Adoption Quarterly. 2009;12(1):37–52. doi: 10.1080/10926750902791953. [DOI] [Google Scholar]
- Elander J, Rutter M. Use and development of the Rutter Parents’ and Teachers’ Scales. International Journal of Methods in Psychiatric Research. 1995;5:1–16. [Google Scholar]
- Elliott A, McMahon C. Anxiety among an Australian sample of young girls adopted from China. Adoption Quarterly. 2011;14(3):161–180. doi: 10.1080/10926755.2011.560792. [DOI] [Google Scholar]
- Ellis BH, Fisher PA, Zaharie S. Predictors of disruptive behavior, developmental delays, anxiety, and affective symptomatology among institutionally reared Romanian children. Journal of the American Academy of Child and Adolescent Psychiatry. 2004;43(10):1283–1292. doi: 10.1097/01.Chi.0000136562.24085.160. [DOI] [PubMed] [Google Scholar]
- Elmund A, Lindblad F, Vinnerljung B, Hjern A. Intercountry adoptees in out-of-home care: a national cohort study. Acta Pædiatrica. 2007;96(3):437–442. doi: 10.1111/j.1651-2227.2006.00149.x. [DOI] [PubMed] [Google Scholar]
- Fisher L, Ames EW, Chisholm K, Savoie L. Problems reported by parents of Romanian orphans adopted to British Colombia. International Journal of Behavioral Development. 1997;20(1):67–82. doi: 10.1080/016502597385441. [DOI] [Google Scholar]
- Gleitman I, Savaya R. Adjustment of adolescent adoptees: The role of age of adoption and exposure to pre-adoption stressors. Children and Youth Services Review. 2011;33(5):758–766. doi: 10.1016/J.Childyouth.2010.11.020. [DOI] [Google Scholar]
- Goldfarb W. The effects of early institutional care on adolescent personality. Journal of Experimental Education. 1943;12:106–129. [Google Scholar]
- Goldfarb W. Infant rearing as a factor in foster home replacement. American Journal of Orthopsychiatry. 1944;14:162–166. [Google Scholar]
- Goldfarb W. Effects of psychological deprivation in infnacy and subsequent stimulation. American Journal of Psychiatry. 1945;102:18–33. [Google Scholar]
- Goldfarb W. Variations in adolescent adjustment of institutionally-reared children. American Journal of Orthopsychiatry. 1947;17:449–457. doi: 10.1111/j.1939-0025.1947.tb05018.x. [DOI] [PubMed] [Google Scholar]
- Goldfarb W. Emotional and intellectual consequences of psychologic deprivation in infancy: A reevaluation. In: Hoch P, Zubin J, editors. Psychopathology of Childhood. New York: Grune & Stratton; 1955. pp. 105–119. [PubMed] [Google Scholar]
- Goldney RD, Donald M, Sawyer MG, Kosky RJ, Priest S. Emotional health of Indonesian adoptees living in Australian families. Australian and New Zealand Journal of Psychiatry. 1996;30(4):534–539. doi: 10.3109/00048679609065029. [DOI] [PubMed] [Google Scholar]
- Greenough WT, Black JE, Wallace CS. Experience and brain development. Child Development. 1987;58(3):539–559. [PubMed] [Google Scholar]
- Gunnar MR. Effects of early deprivation: Findings from orphanage-reared infants and children. In: Nelson CA, Luciana M, editors. Handbook of Developmental Cognitive Neuroscience. Cambridge, MA: MIT Press; 2001. pp. 617–629. [Google Scholar]
- Gunnar MR. Integrating neuroscience and psychological approaches in the study of early experience. Annals of the New York Academy of Sciences. 2003;1008:238–247. doi: 10.1196/annals.1301.024. [DOI] [PubMed] [Google Scholar]
- Gunnar MR, Fisher PA, The Early Experience, Stress and Prevention Network Bringing basic research on early experience and stress neurobiology to bear on preventive interventions for neglected and maltreated children. Development and Psychopathology. 2006;18(3):651–677. [PubMed] [Google Scholar]
- Gunnar MR, Van Dulmen MHM, The International Adoption Project Team Behavior problems in postinstitutionalized internationally adopted children. Development and Psychopathology. 2007;19(1):129–148. doi: 10.1017/S0954579407070071. [DOI] [PubMed] [Google Scholar]
- Habersaat SA, Tessier R, Pierrehumbert B. International adoption: Influence of attachment and maternal monitoring style in the emergence of behavioural problems in adolescence in relation to age at adoption. Schweizer Archiv für Neurologie Und Psychiatrie. 2011;162(1):21–26. [Google Scholar]
- Harlow HF. The nature of love. American Psychologist. 1958;13:673–685. doi: 10.1037/h0029383. [DOI] [PubMed] [Google Scholar]
- Harlow HF, Novak MA. Psychopathological perspectives. Perspectives in Biology and Medicine. 1973;16(3):461–478. doi: 10.1353/pbm.1973.0022. [DOI] [PubMed] [Google Scholar]
- Hawk BN, McCall RB. Specific extreme behaviors of postinstitutionalized Russian adoptees. Developmental Psychology. 2011;47(3):732–738. doi: 10.1037/a0021108. [DOI] [PubMed] [Google Scholar]
- Jensen AR. Cumulative deficit: A testable hypothesis? Developmental Psychology. 1974;10(6):996–1019. [Google Scholar]
- Juffer F, van IJzendoorn MH. Behavior problems and mental health referrals of international adoptees: A meta-analysis. Journal of the American Medical Association. 2005;293(20):2501–2515. doi: 10.1001/jama.293.20.2501. [DOI] [PubMed] [Google Scholar]
- Kertes DA, Gunnar MR, Madsen NJ, Long JD. Early deprivation and home basal cortisol levels: A study of internationally adopted children. Development and Psychopathology. 2008;20:473–491. doi: 10.1017/S0954579408000230. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Knudsen EI. Sensitive periods in the development of brain and behavior. Journal of Cognitive Neuroscience. 2004;16(8):1412–1425. doi: 10.1162/0898929042304796. [DOI] [PubMed] [Google Scholar]
- Kraemer GW. A psychobiological theory of attachment. Behavioral and Brain Sciences. 1992;15(3):493–541. doi: 10.1017/S0140525X00069752. [DOI] [PubMed] [Google Scholar]
- Kreppner JM, O’Connor TG, Rutter M, The English and Romanian Adoption Study Team Can inattention/overactivity be an institutional deprivation syndrome? Journal of Abnormal Child Psychology. 2001;29(6):513–528. doi: 10.1023/a:1012229209190. [DOI] [PubMed] [Google Scholar]
- Kreppner JM, Rutter M, Beckett C, Castle J, Colvert E, Groothues C, Sonuga-Barke EJS. Normality and impairment following profound early institutional deprivation: A longitudinal follow-up into early adolescence. Developmental Psychology. 2007;43(4):931–946. doi: 10.1037/0012-1649.43.4.931. [DOI] [PubMed] [Google Scholar]
- Kvifte Andresen IL. Behavioural and school adjustment of 12–13-year old internationally adopted children in Norway: A research note. Journal of Child Psychology and Psychiatry. 1992;33(2):427–439. doi: 10.1111/j.1469-7610.1992.tb00877.x. [DOI] [PubMed] [Google Scholar]
- Lee RM, Seol KO, Sung M, Miller MJ, Minnesota International Adoption Project Team The behavioral development of Korean children in institutional care and international adoptive families. Developmental Psychology. 2010;46(2):468–478. doi: 10.1037/a0017358. [DOI] [PubMed] [Google Scholar]
- Lindblad F, Weitoft GR, Hjern A. ADHD in international adoptees: A national cohort study. European Child and Adolescent Psychiatry. 2010;19(1):37–44. doi: 10.1007/s00787-009-0038-3. [DOI] [PubMed] [Google Scholar]
- MacLean K. The impact of institutionalization on child development. Development and Psychopathology. 2003;15(4):853–884. doi: 10.1017/s0954579403000415. [DOI] [PubMed] [Google Scholar]
- Marcovitch S, Cesaroni L, Roberts W, Swanson C. Romanian adoption: Parents’ dreams, nightmares, and realities. Child Welfare. 1995;74(5):993–1017. [Google Scholar]
- Marcovitch S, Goldberg S, Gold A, Washington J, Wasson C, Krekewich K, Handley-Derry M. Determinants of behavioural problems in Romanian children adopted to Ontario. International Journal of Behavioral Development. 1997;20(1):17–31. [Google Scholar]
- Maurer D, Mondloch CJ, Lewis TI. Sleeper effects. Developmental Science. 2007;10(1):40–47. doi: 10.1111/j.1467-7687.2007.00562.x. [DOI] [PubMed] [Google Scholar]
- McCall RB. IX. Research, practice, and policy perspectives on issues of children without permanent parental care. Monographs of the Society for Research in Child Development. 2011;76(4):223–272. doi: 10.1111/j.1540-5834.2011.00634.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Meaney MJ. Epigenetics and the biological definition of gene x environment interactions. Child Development. 2010;81(1):41–79. doi: 10.1111/j.1467-8624.2009.01381.x. [DOI] [PubMed] [Google Scholar]
- Mersky JP, Topitzes JD, Reynolds AJ. Maltreatment prevention through early childhood intervention: A confirmatory evaluation of the Chicago Child-Parent Center preschool program. Children and Youth Services Review. 2011;33:1454–1463. doi: 10.1016/j.childyouth.2011.04.022. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Merz EC, McCall RB. Behavior problems in children adopted from psychosocially depriving institutions. Journal of Abnormal Child Psychology. 2010;38(4):459–470. doi: 10.1007/s10802-009-9383-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Merz EC, McCall RB. Parent ratings of executive functioning in children adopted from psychosocially depriving institutions. Journal of Child Psychology and Psychiatry. 2011;52(5):537–546. doi: 10.1111/j.1469-7610.2010.02335.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Miller L, Chan W, Tirella L, Perrin E. Outcomes of children adopted from Eastern Europe. International Journal of Behavioral Development. 2009;33(4):289–298. doi: 10.1177/0165025408098026. [DOI] [Google Scholar]
- National Scientific Council on the Developing Child. Excessive Stress Disrupts the Architecture of the Developing Brain: Working Paper #3. 2005 http://www.developingchild.net.
- National Scientific Council on the Developing Child. Mental Health Problems in Early Childhood Can Impair Learning and Behavior for Life: Working Paper #6. 2008 http://www.developingchild.net.
- National Scientific Council on the Developing Child. Persistent Fear and Anxiety Can Affect Young Children’s Learning and Development: Working Paper 9. 2010 http://www.developingchild.net.
- National Scientific Council on the Developing Child. Building the Brain’s “Air Traffic Control” System: How Early Experiences Shape the Development of Executive Function: Working Paper No. 11. 2011 http://www.developingchild.net.
- Nielsen A, Coleman PK, Guinn M, Robb C. Length of institutionalization, contact with relatives, and previous hospitalizations as predictors of social and emotional behavior in young Ugandan orphans. Childhood. 2004;11(1):94–116. doi: 10.1177/0907568204037116. [DOI] [Google Scholar]
- Niemann S, Weiss S. Factors affecting attachment in international adoptees at 6 months post adoption. Children and Youth Services Review. 2012;34:205–212. doi: 10.1016/j.childyouth.2011.10.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Novak MA, Harlow HF. Social recovery of monkeys isolated for the first year of life: I. Rehabilitation and therapy. Developmental Psychology. 1975;11:453–465. [Google Scholar]
- O’Connor TG. Early experiences and psychological development: Conceptual questions, empirical illustrations, and implications for intervention. Development and Psychopathology. 2003;15(3):671–690. doi: 10.1017/s0954579403000336. doi: 10.1017.S0954579403000336. [DOI] [PubMed] [Google Scholar]
- O’Connor TG, Bredenkamp D, Rutter M, The English and Romanian Adoptees (ERA) Study Team Attachment disturbances and disorders in children exposed to early severe deprivation. Infant Mental Health Journal. 1999;20(1):10–29. [Google Scholar]
- O’Connor TG, Marvin RS, Rutter M, Olrick JT, Britner PA, The English and Romanian Adoptees Study Team Child-parent attachment following early institutional deprivation. Development and Psychopathology. 2003;15(1):19–38. doi: 10.1017/s0954579403000026. doi: 10.1017.S0954579403000026. [DOI] [PubMed] [Google Scholar]
- O’Connor TG, Rutter M, The English and Romanian Adoptees (ERA) Study Team Attachment disorder behavior following early severe deprivation: Extension and longitudinal follow-up. Journal of the American Academy of Child and Adolescent Psychiatry. 2000;39(6):703–712. doi: 10.1097/00004583-200006000-00008. [DOI] [PubMed] [Google Scholar]
- Olds DL, Eckenrode J, Henderson CR, Jr, Kitzman H, Powers J, Cole R, Luckey D. Long-term effects of home visitation on maternal life course and child abuse and neglect. Fifteen-year follow-up of a randomized trial. Journal of the American Medical Association. 1997;278(8):637–643. [PubMed] [Google Scholar]
- Parker KJ, Maestripieri D. Identifying key features of early stressful experiences that produce stress vulnerability and resilience in primates. Neuroscience and Biobehavioral Reviews. 2011;35(7):1466–1483. doi: 10.1016/j.neubiorev.2010.09.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Reynolds AJ, Robertson D. School-based early intervention and later child maltreatment in the Chicago Longitudinal Study. Child Development. 2003;74(1):3–26. doi: 10.1111/1467-8624.00518. [DOI] [PubMed] [Google Scholar]
- Rojewski JW, Shapiro MS, Shapiro M. Parental assessment of behavior in Chinese adoptees during early childhood. Child Psychiatry and Human Development. 2000;31(1):79–96. doi: 10.1023/a:1001978122206. [DOI] [PubMed] [Google Scholar]
- Rutter M. Maternal deprivation reassessed. second. Harmondsworth, UK: Penguin Books; 1981. [Google Scholar]
- Rutter M. Genetic influences on risk and protection: Implications for understanding resilience. In: Luthar S, editor. Resilience and vulnerability: Adaptation in the context of childhood adversities. New York: Cambridge University Press; 2003. pp. 489–509. [Google Scholar]
- Rutter M. Adverse preadoption experiences and psychological outcomes. In: Brodzinsky D, Palacios J, editors. Psychological issues in adoption: Research and practice. Westport, Conn: Praeger Publishers; 2005a. pp. 67–92. [Google Scholar]
- Rutter M. The psychological effects of early institutional rearing. In: Marshall PJ, Fox NA, editors. Development of Social Engagement: Neurobiologial Perspectives. Cary, NC, USA: Oxford University Press; 2005b. pp. 355–391. [Google Scholar]
- Rutter M, Colvert E, Kreppner J, Beckett C, Castle J, Groothues C, Sonuga-Barke EJS. Early adolescent outcomes for institutionally-deprived and non-deprived adoptees. I: Disinhibited attachment. Journal of Child Psychology and Psychiatry. 2007;48(1):17–30. doi: 10.1111/j.1469-7610.2006.01688.x. [DOI] [PubMed] [Google Scholar]
- Rutter M, Kreppner JM, O’Connor TG. Specificity and heterogeneity in children’s responses to profound institutional privation. British Journal of Psychiatry. 2001;179:97–103. doi: 10.1192/bjp.179.2.97. [DOI] [PubMed] [Google Scholar]
- Rutter M, Moffit T, Caspi A. Gene-environment interplay and psychopathology: Multiple varieties but real effects. Journal of Child Psychology and Psychiatry. 2006;47:226–261. doi: 10.1111/j.1469-7610.2005.01557.x. [DOI] [PubMed] [Google Scholar]
- Rutter M, O’Connor TG, The English and Romanian Adoptees (ERA) Study Team Are there biological programming effects for psychological development? Findings from a study of Romanian adoptees. Developmental Psychology. 2004;40(1):81–94. doi: 10.1037/0012-1649.40.1.81. [DOI] [PubMed] [Google Scholar]
- Rutter M, Sonuga-Barke EJ, Beckett C, Castle J, Kreppner J, Kumsta R, Bell CA. Deprivation-specific psychological patterns: Effects of institutional deprivation. Monographs of the Society for Research in Child Development. 2010;75(1):1–250. doi: 10.1111/j.1540-5834.2010.00550.x. [DOI] [PubMed] [Google Scholar]
- Schaffer HR. The early experience assumption: Past, present, and future. International Journal of Behavioral Development. 2000;24(1):5–14. doi: 10.1080/016502500383412. [DOI] [Google Scholar]
- Shonkoff JP, Boyce WT, McEwen BS. Neuroscience, molecular biology, and the childhood roots of health disparities: Building a new framework for health promotion and disease prevention. Journal of the American Medical Association. 2009;301(21):2252–2259. doi: 10.1001/jama.2009.754. [DOI] [PubMed] [Google Scholar]
- Smyke AT, Koga SF, Johnson DE, Fox NA, Marshall PJ, Nelson CA, The BEIP Core Group The caregiving context in institution-reared and family-reared infants and toddlers in Romania. Journal of Child Psychology and Psychiatry. 2007;48(2):210–218. doi: 10.1111/j.1469-7610.2006.01694.x. [DOI] [PubMed] [Google Scholar]
- Smyke AT, Zeanah CH, Fox NA, Nelson CA, Guthrie D. Placement in foster care enhances quality of attachment among young institutionalized children. Child Development. 2010;81(1):212–223. doi: 10.1111/j.1467-8624.2009.01390.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Smyke AT, Zeanah CH, Gleason MM, Drury SS, Fox NA, Nelson CA, Guthrie D. A randomized controlled trial comparing foster care and institutional care for children with signs of Reactive Attachment Disorder. American Journal of Psychiatry. 2012;169(5):508–514. doi: 10.1176/appi.ajp.2011.11050748. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Sonuga-Barke EJS, Beckett C, Kreppner J, Castle J, Colvert E, Stevens S, Rutter M. Is subnutrition necessary for a poor outcome following early institutional deprivation? Developmental Medicine & Child Neurology. 2008;50(9):664–671. doi: 10.1111/j.1469-8749.2008.03065.x. [DOI] [PubMed] [Google Scholar]
- Spitz R. Hospitalism: A follow-up report. Psychoanalytic Study of the Child. 1945a;2:113–118. [PubMed] [Google Scholar]
- Spitz R. Hospitalism: An inquiry into the genesis of psychiatric conditions in early childhood. Psychoanalytic Study of the Child. 1945b;1:53–74. [PubMed] [Google Scholar]
- Sroufe LA, Waters E. Attachment as an organizational construct. Child Development. 1977;48:1184–1199. [Google Scholar]
- Stevens S, Sonuga-Barke EJS, Kreppner J, Beckett C, Castle J, Colvert E, Rutter M. Inattention/overactivity following early severe institutional deprivation: Presentation and associations in early adolescence. Journal of Abnormal Child Psychology. 2008;36(3):385–398. doi: 10.1007/s10802-007-9185-5. [DOI] [PubMed] [Google Scholar]
- Suomi SJ. Gene-environment interactions and the neurobiology of social conflict. Annals of the New York Academy of Sciences. 2003;1008:132–139. doi: 10.1196/annals.1301.014. [DOI] [PubMed] [Google Scholar]
- Tan TX. History of early neglect and middle childhood social competence: An adoption study. Adoption Quarterly. 2006;9(4):59–72. doi: 10.1300/J145v09n4_04. [DOI] [Google Scholar]
- Tan TX, Camras LA. Social skills of adopted Chinese girls at home and in school: Parent and teacher ratings. Children and Youth Services Review. 2011;33:1813–1821. doi: 10.1016/j.childyouth.2011.05.006. [DOI] [Google Scholar]
- Tan TX, Camras LA, Deng H, Zhang M, Lu Z. Family stress, parenting styles, and behavioral adjustment in preschool-age adopted Chinese girls. Early Childhood Research Quarterly. 2012;27:128–136. doi: 10.1016/j.ecresq.2011.04.002. [DOI] [Google Scholar]
- Tan TX, Marfo K. Parental ratings of behavioral adjustment in two samples of adopted Chinese girls: Age-related versus socio-emotional correlates and predictors. Journal of Applied Developmental Psychology. 2006;27(1):14–30. doi: 10.1016/j.appdev.2005.12.004. [DOI] [Google Scholar]
- Tan TX, Marfo K, Dedrick RF. Special needs adoption from China: Exploring child-level indicators, adoptive family characteristics, and correlates of behavioral adjustment. Children and Youth Services Review. 2007;29(10):1269–1285. doi: 10.1016/j.childyouth.2007.05.001. [DOI] [Google Scholar]
- Tan TX, Marfo K, Dedrick RF. Early developmental and psychosocial risks and longitudinal behavioral adjustment outcomes for preschool-age girls adopted from China. Journal of Applied Developmental Psychology. 2010;31:306–314. doi: 10.1016/j.appdev/2010.04.002. [DOI] [Google Scholar]
- The St. Petersburg-USA Orphanage Research Team. The effects of early social-emotional and relationship experience on the development of young orphanage children. Monographs of the Society for Research in Child Development. 2008;291, 73(3) doi: 10.1111/j.1540-5834.2008.00483.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Tizard B, Hodges J. The effect of early institutional rearing on the development of eight year old children. Journal of Child Psychology and Psychiatry. 1978;19(2):99–118. doi: 10.1111/j.1469-7610.1978.tb00453.x. [DOI] [PubMed] [Google Scholar]
- Tizard B, Rees J. The effect of early institutional rearing on the behaviour problems and affectional relationships of four-year-old children. Journal of Child Psychology and Psychiatry. 1975;16(1):61–73. doi: 10.1111/j.1469-7610.1975.tb01872.x. [DOI] [PubMed] [Google Scholar]
- van den Dries L, Juffer F, van IJzendoorn MH, Bakermans-Kranenburg MJ. Fostering security? A meta-analysis of attachment in adopted children. Children and Youth Services Review. 2009;31(3):410–421. doi: 10.1016/j.childyouth.2008.09.008. [DOI] [Google Scholar]
- van IJzendoorn MH, Bakermans-Kranenburg MJ, Juffer F. Plasticity of growth in height, weight and head circumference: Meta-analytic evidence for massive catch-up after international adoption. Journal of Developmental and Behavioral Pediatrics. 2007;28(4):334–343. doi: 10.1097/DBP.0b013e31811320aa. [DOI] [PubMed] [Google Scholar]
- van IJzendoorn MH, Juffer F. The Emanuel Miller Memorial Lecture 2006: Adoption as intervention. Meta-analytic evidence for massive catch-up and plasticity in physical, socio-emotional, and cognitive development. Journal of Child Psychology and Psychiatry. 2006;47(12):1228–1245. doi: 10.1111/j.1469-7610.2006.01675.x. [DOI] [PubMed] [Google Scholar]
- van IJzendoorn MH, Palacios J, Sonuga-Barke EJ, Gunnar MR, Vorria P, McCall RB, Juffer F. I. Children in institutional care: Delayed development and resilience. Monographs of the Society for Research in Child Development. 2011;76(4):8–30. doi: 10.1111/j.1540-5834.2011.00626.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Verhulst FC, Althaus M, Versluis-Den Bieman HJM. Problem behavior in international adoptees: II. Age at placement. Journal of the American Academy of Child and Adolescent Psychiatry. 1990;29(1):104–111. doi: 10.1097/00004583-199001000-00016. [DOI] [PubMed] [Google Scholar]
- Verhulst FC, Althaus M, Versluis-Den Bieman HJM. Damaging backgrounds: Later adjustment of international adoptees. Journal of the American Academy of Child and Adolescent Psychiatry. 1992;31(3):518–524. doi: 10.1097/00004583-199205000-00020. [DOI] [PubMed] [Google Scholar]
- Vorria P, Rutter M, Pickles A, Wolkind S, Hobsbaum A. A comparative study of Greek children in long-term residential group care and in two-parent families: I. Social, emotional, and behavioural differences. Journal of Child Psychology and Psychiatry. 1998a;39(2):225–236. [PubMed] [Google Scholar]
- Vorria P, Rutter M, Pickles A, Wolkind S, Hobsbaum A. A comparative study of Greek children in long-term residential group care and in two-parent families: II. Possible mediating mechanisms. Journal of Child Psychology and Psychiatry. 1998b;39(2):237–245. [PubMed] [Google Scholar]
- Wolkind SN. The components of “affectionless psychopathy” in institutionalized children. Journal of Child Psychology and Psychiatry. 1974;15:215–220. doi: 10.1111/j.1469-7610.1974.tb01246.x. [DOI] [PubMed] [Google Scholar]
- Zeanah CH, Gunnar MR, McCall RB, Kreppner J, Fox NA. VI. Sensitive periods. Monographs of the Society for Research in Child Development. 2011;76(4):147–162. doi: 10.1111/j.1540-5834.2011.00631.x. [DOI] [PMC free article] [PubMed] [Google Scholar]