19
This article was downloaded by: [University of Washington Libraries] On: 19 November 2014, At: 01:22 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK Adoption Quarterly Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/wado20 A Study of Adopted Children, Their Environment, and Development: A Systematic Review Mogens Nygaard Christoffersen a a The Danish National Institute of Social Research , Copenhagen , Denmark Published online: 04 Sep 2012. To cite this article: Mogens Nygaard Christoffersen (2012) A Study of Adopted Children, Their Environment, and Development: A Systematic Review, Adoption Quarterly, 15:3, 220-237, DOI: 10.1080/10926755.2012.700002 To link to this article: http://dx.doi.org/10.1080/10926755.2012.700002 PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms- and-conditions

A Study of Adopted Children, Their Environment, and Development: A Systematic Review

Embed Size (px)

Citation preview

Page 1: A Study of Adopted Children, Their Environment, and Development: A Systematic Review

This article was downloaded by: [University of Washington Libraries]On: 19 November 2014, At: 01:22Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Adoption QuarterlyPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/wado20

A Study of Adopted Children, TheirEnvironment, and Development: ASystematic ReviewMogens Nygaard Christoffersen aa The Danish National Institute of Social Research , Copenhagen ,DenmarkPublished online: 04 Sep 2012.

To cite this article: Mogens Nygaard Christoffersen (2012) A Study of Adopted Children, TheirEnvironment, and Development: A Systematic Review, Adoption Quarterly, 15:3, 220-237, DOI:10.1080/10926755.2012.700002

To link to this article: http://dx.doi.org/10.1080/10926755.2012.700002

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the“Content”) contained in the publications on our platform. However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoever as tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand views expressed in this publication are the opinions and views of the authors,and are not the views of or endorsed by Taylor & Francis. The accuracy of the Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoever orhowsoever caused arising directly or indirectly in connection with, in relation to or arisingout of the use of the Content.

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Page 2: A Study of Adopted Children, Their Environment, and Development: A Systematic Review

Adoption Quarterly, 15:220–237, 2012Copyright © Taylor & Francis Group, LLCISSN: 1092-6755 print / 1544-452X onlineDOI: 10.1080/10926755.2012.700002

A Study of Adopted Children, TheirEnvironment, and Development: A Systematic

Review

MOGENS NYGAARD CHRISTOFFERSENThe Danish National Institute of Social Research, Copenhagen, Denmark

The study explores the developmental consequences for childrenwho have been adopted. The comparison group is comprised oftheir non-adopted siblings or children in foster homes or groupcare. A search in scientific databases resulted in more than 3,300hits about adoption. Among these, 56 fulfilled the selection crite-ria from 17 studies across more than 2,000 adoptees and con-trols. Adopted children scored higher on IQ, school-performance,and lack of behavioral problems than their non-adopted siblings orpeers who stayed behind in orphanages or foster homes. The resultsfrom OECD countries were similar to those from studies in Chile,Lebanon, and India.

KEYWORDS adoption, behavior problems, self-esteem, IQ, foster-ing, mental health

INTRODUCTION

Adoption has a long history as a form of care for children in need, andadopted children generally show a good developmental outcome. Researchmethodology used in studies about adoption often gives rise to criticismrelated to the lack of an adequate comparison group, small samples, ret-rospective methods, etc. (Cohen, 2002). Most studies compare the adopted

Received 31 August 2011; revised 23 February 2012; accepted 23 March 2012.An earlier version of this article was presented at the 16th Nordic Demographic Sympo-

sium, Helsinki, June 5–7, 2008. I would like to thank the Ministry of Social affairs for grantingthe review. I owe a debt of gratitude to Ida Hammen, Karen Raft Andersen, and NadiaJeldtoft for their assistance in reviewing the references. Jesus Palacios was extremely helpfulin providing additional information about his and Y. Sanchez-Sandoval’s research project.

Address correspondence to Mogens Nygaard Christoffersen, The Danish National Instituteof Social Research, Herluf Trollesgade 11, Copenhagen DK-1052, Denmark. E-mail: [email protected]

220

Dow

nloa

ded

by [

Uni

vers

ity o

f W

ashi

ngto

n L

ibra

ries

] at

01:

22 1

9 N

ovem

ber

2014

Page 3: A Study of Adopted Children, Their Environment, and Development: A Systematic Review

Adopted Children, Their Environment, and Development 221

children with their new classmates or children in the new environment. In-ternationally adopted children who cross borders are compared to matchedsamples of children in the new country or intercountry adopted children arecompared to new siblings or classmates in their new environment. Thesestudies are unable to answer the question: What would have happened tothese children if they had not been adopted?

In the present study, the comparison group is comprised of the childrenleft in orphanages or in foster homes or with their non-adopted siblings.These comparison groups allow a better opportunity to control for the con-sequences of early adverse experiences. Furthermore, the present surveycan take advantage of the results from many small studies, thus reducing theinsecurity caused by the small sample size.

Previous Systematic Reviews of Adoptions

One systematic review of the physical development of internationallyadopted children shows that the delay in physical growth, height, and weightdepends on the number of months spent in an institution before adoption.In particular, children adopted after their first birthday show a delay in theirphysical development. The review covers more than 120 studies publishedsince 1956. The comparison groups are the children in the new country, andthe results from the review show that the internationally adopted childrengenerally have a lower height than their new contemporaries (van IJzen-doorn, Bakermans-Kranenburg, & Juffer, 2007). This review fails to answerthe question of what would have happened if these children had not beenadopted, because their height is compared to children in the new environ-ment instead of the children who have been left in institutional care etc.

Another systematic study of 62 studies covering about 17,000 childrencompares the IQ of internationally adopted children with children in the newcountry. In this review the comparison groups were living in an environmentcomparable to the adoptive families. The results show that the adopted chil-dren catch up on cognitive development. On average there is no significantdifference between the IQ of the adopted and the control groups (van IJzen-doorn, Juffer, & Poelhuis, 2005). The question is whether these comparisonsunderestimate the adopted children’s gains by being adopted instead of liv-ing in foster care or in a group care arrangement.

Adopted children may suffer from consequences of neglect, abuse, andmalnutrition in institutions before adoption. They have to cope with theiradoptive status and possibly a lack of resemblance to their adoptive parents,leading to an increased risk of low self-esteem. Contrary to these expecta-tions, a meta-analysis of studies on self-esteem of transracial, international,and domestic adoptees found no differences between adoptees and theirnon-adopted peers across 88 studies with 10,000 adoptees and 33,000 non-adopted children. Across 18 studies, the reviewers found no differences

Dow

nloa

ded

by [

Uni

vers

ity o

f W

ashi

ngto

n L

ibra

ries

] at

01:

22 1

9 N

ovem

ber

2014

Page 4: A Study of Adopted Children, Their Environment, and Development: A Systematic Review

222 M. N. Christoffersen

in self-esteem between transracial and same-race adoptions (Juffer & vanIJzendoorn, 2007).

A systematic review of the behavioral and mental problems experiencedby internationally adopted children compares their problems with their con-temporaries in the new environment. The researchers found 34 studies aboutmental health problems and 64 studies about behavioral problems coveringabout 25,000 adopted children and 80,000 children in comparison groups.The adoptees were significantly overrepresented in mental health services.The review showed that the adopted children had more behavioral problemsthan those in the control groups, but the differences were small (Juffer &van IJzendoorn, 2005).

In a specific analysis of studies of adopted children tested in the agegroup between 12 and 20 years and compared to samples of contempo-raries in the new country, Bimmel, Juffer, van IJzendoorn, and Bakermans-Kranenburg (2003) found 10 studies with 2,300 adopted children and 14,000children in the comparison groups. The meta-analysis showed that inter-nationally adopted children exhibit more externalizing behavioral problemsthan non-adopted adolescents, but the differences were small. An explana-tion for these findings may be the adoptees’ previous experiences. Beforeadoption, some of the internationally adopted children experience insuffi-cient medical care, malnutrition, neglect, and abuse in orphanages (Juffer &van IJzendoorn, 2005).

The previous systematic reviews have focused on adopted children andtheir developmental progress compared to their new peers. That is the reasonthe previous reviews have difficulty in answering the question: What wouldhave happened to these children if they had not been adopted? Have thechildren benefited from adoption when compared to those children whohave been left in a foster home or group care arrangement, orphanage, orinstitutions? To capture the gains and losses of adoption, these questionsmust be addressed.

Method

The purpose of the present study is to study adoption as a protective factorfor children and adolescents who otherwise would grow up in care arrange-ments because their parents are not capable of taking care of them, giventheir circumstances and according to contemporary social support systems(Christoffersen, Hammen, Andersen, & Jeldtoft, 2007).

Inclusion and Exclusion Criteria

The review includes only studies that compare outcome for adopted childrenwith outcome for children who are not adopted, such as siblings of adoptedchildren or children who grow up in foster families, in group care, or ininstitutions. International adoption across borders will normally be excluded

Dow

nloa

ded

by [

Uni

vers

ity o

f W

ashi

ngto

n L

ibra

ries

] at

01:

22 1

9 N

ovem

ber

2014

Page 5: A Study of Adopted Children, Their Environment, and Development: A Systematic Review

Adopted Children, Their Environment, and Development 223

because these studies lack a comparison group of adolescents growing upunder the circumstances that the adopted adolescents have left.1 Likewise,children adopted by their stepparents will not be included.

The review includes adoptions used as an intervention in families whowere not able to take care of their children. The adoption could have takenplace with or without consent, it could be anonymous, or it could be openwith some contact between the family of origin and the adoptive parents.

Private adoptions are not included, and only children known to thepublic authorities are included in the present review. There are no agelimits: the children might have been adopted as newborns, from 1 to 2 yearsold, or at a much older age.

Outcomes

The review of the developmental consequences of adoption includes a num-ber of outcome variables which indicate the physical growth, the cognitivedevelopment (e.g., IQ and school achievements), social and emotional de-velopment (e.g., self-esteem, mental health problems, and behavioral prob-lems).

Search Strategy for Identification of Studies

In order to minimize publication bias, the overall search strategy in thisstudy was designed to identify unpublished research (i.e., conference papers,unpublished dissertations, or government reports including a keyword searchof computerized databases). We have contacted authors of included studiesand also other researchers working in this area asking about their knowledgeof ongoing research programs and unpublished papers and working papers.

Several strategies have been used to identify studies, published or oth-erwise, that meet the criteria, including a keyword search of computerizeddatabases and examination of electronic databases. References from recentsystematic or traditional reviews of adoption were hand-searched.

The search resulted in more than 3,300 hits about adoption. The title,abstracts, and if necessary also the books and articles were searched. Theseries of databases were browsed with a search string including the followingterms: Adoptee, Adopted, Adoption, Permanent Planning, Child.2 The 3,300hits were entered and organized in Reference Manager.

A data extraction coding scheme was developed for this review withinformation on study design, intervention characteristics, and control groups’characteristics. The standardized scheme was developed to categorize thestudies which fulfilled the inclusion criteria.

Titles and abstracts of studies identified through the searches of elec-tronic databases were reviewed to determine whether they met the inclusioncriteria. If there was uncertainty or disagreement, the studies were discussed

Dow

nloa

ded

by [

Uni

vers

ity o

f W

ashi

ngto

n L

ibra

ries

] at

01:

22 1

9 N

ovem

ber

2014

Page 6: A Study of Adopted Children, Their Environment, and Development: A Systematic Review

224 M. N. Christoffersen

among the reviewers. Relevant and potentially relevant articles were retrievedand assessed independently against the inclusion criteria. If there was uncer-tainty or disagreement was unresolved, additional information was soughtfrom the authors.

The following databases were used:

• MEDLINE/PubMed• Embase• Social Science Citation Index• SocINDEX• PsycINFO

The search resulted in 56 hits which fulfilled the selection criteria from17 studies across 1,200 adoptees and 1,200 non-adopted comparisons. Thestudies were published in the years 1977 to 2006. They were conducted inOECD countries such as Canada, Denmark, England, France, New Zealand,Spain, Sweden, and the United States as well as Chile, Lebanon, and India.

Data Synthesis and Analysis

Data synthesis was conducted within RevMan 5, the latest version of theCochrane Collaboration’s meta-analysis software (RevMan, 2008). When aprimary outcome study reports multiple measures of the same construct (i.e.,IQ or school achievements) at different points in time, we used the singlemeasure that is newest as an overall measure because we were looking forthe long-term consequences.

CONTINUOUS DATA

Continuous data were analyzed if means and standard deviations wereavailable or obtained from primary investigators or otherwise derived. TheRevMan formula for standardized mean difference (SMD) is Hedges’ g, whichis like Cohen’s d but includes an adjustment for small sample bias (RevMan,2008). Inverse variance method was used to pool SMDs, so that each effectsize was weighted by the inverse of its variance in an overall estimate ofeffect size. Confidence intervals of 95% were used for individual study dataand pooled estimates.

BINARY DATA

Binary outcomes were analyzed by calculating odds ratios with 95% confi-dence intervals. If some primary studies reported an outcome as a dichoto-mous measure and others used a continuous measure of the same construct,

Dow

nloa

ded

by [

Uni

vers

ity o

f W

ashi

ngto

n L

ibra

ries

] at

01:

22 1

9 N

ovem

ber

2014

Page 7: A Study of Adopted Children, Their Environment, and Development: A Systematic Review

Adopted Children, Their Environment, and Development 225

a conversion between the two metrics was performed (Chinn, 2000; Lipsey& Wilson, 2001).

Assessment of Heterogeneity

It is a condition for the meta-analysis that the group of trials was sufficientlyhomogenous in terms of participants, intervention and outcomes.

The variability among studies in a systematic review is termed hetero-geneity. This can arise from methodological diversity (e.g., studies may sufferfrom different degrees of bias) or differences in outcome assessments (e.g.,not all studies estimate the same construct). If substantial heterogeneity wasfound in the primary study, the following factors were considered as possi-ble explanations: intervention type (age when adopted) and characteristics oftype of research design (age when tested, type of measurements of construct,missing information, time of publication).

Heterogeneity is evaluated in three ways: (1) with I2, (2) with the chi-square test of heterogeneity test, or (3) by comparing results of fixed andrandom effects models (Higgins, 2002). Possible sources of heterogeneitywere investigated. If there was evidence of heterogeneity (i.e., the statisticaltest of heterogeneity is statistically significant and I2 >50% (Deeks, Higgins,& Altman, 2006), both fixed effect and random effects models were examinedand the random model considered.

A random effect model estimates the total variance associated with thedistribution of effect size as a sum of the estimate of the variance betweenstudies and the estimate of the variance associated with subject level sam-pling error (Lipsey & Wilson, 2001).

RESULTS

Cognitive Development

Six studies met the inclusion criteria for exploring the cognitive developmentof adopted children. The studies compared the developmental gains foradopted children with children who remained in, for example, a group carearrangement or an institution, or who stayed with a foster family or witha sibling (Table 1). Two studies were lacking information about standarddeviation. Additional information was sought from the authors but with nosuccess as yet. Both these studies gave significant results if we assume thatthe standard deviation of the IQ measures would have been on the samelevel as the other studies.

Figure 1 shows the results when these studies were excluded fromthe meta-analysis. The remaining four studies show an individual significantresult in favor of the adopted. The mean difference was estimated to be17.12 (CI, 11.19–23.06) in a random effects model.

Dow

nloa

ded

by [

Uni

vers

ity o

f W

ashi

ngto

n L

ibra

ries

] at

01:

22 1

9 N

ovem

ber

2014

Page 8: A Study of Adopted Children, Their Environment, and Development: A Systematic Review

226 M. N. Christoffersen

TABLE 1 IQ for Adopted vs. Control Children

Country(N )

Age WhenAdopted

Age WhenTested

IQ TestAverage SD

Effect SizeHedges’ g

(Bharat, 1997)a India Before 1 year 8–11 years ICITAdopted (41) 104.33Institution (24) 78.83(Dennis, 1973) Lebanon Varied 12 years

or olderS-B

Adopted, Boys (31) 89.03 16.28Institution, Boys (16) 80.6 13.5 0.54Adopted, Girls (54) 83.02 15.83Institution, Girls (35) 53.7∗∗∗ 7.8 2.19(Hodges &

Tizard, 1989)England 2–4 years 16 years WAIS

Adopted (11) 114.1 11.21Institution (5) 95.8∗∗ 9.8 2.02Biological Parents (8) 96.3∗∗∗ 16.2 1.41(Colombo,

Delaparra, &Lopez, 1992)

Chile Youngerthan3 years

6–12 years WISC

Adopted (16) 96.7 11.7Institution (8) 81.1∗∗ 9.8 1.35Biological Parents (11) 83.3∗ 11.2 1.13(Bagley & Young,

1979)England Younger

than3 years

6–8 years EPVTI

Adopted, MixedEthnicity

(30) 113.9

Adopted, Majority (30) 111.9In Care, Mixed

ethnicity(24) 92.1

(Schiff, 1978) France Youngerthan 6months

6–14 years ECNI

Adopted (32) 106.9 12.2Remained Siblings (20) 95.4∗∗ 12.9 0.91

Note. EPVTI = English version of Picture Vocabulary Test; ECNI = Echelle Collective de Niveau Intel-lectuel; ICIT = Indian Child Intelligence Test; S-B = Stanford-Binet; WAIS = Wechsler Adult IntelligenceScale; WISC = Wechsler Intelligence Scale. The groups of adopted children are used as a referencegroup. A positive effect size measures the difference between the adopted and controls in favor of theadopted. In the calculations of effect size Hedges’ corrections are usedb (cf. Hedges, 1981). In calculationof the pooled effect size the reciprocal variance is used as weights. Effect size is calculated for studieswith information about variance or standard deviation (SD). Also the skipped studies of the adopted hadhigher IQ measures than the controls.∗p < .05; ∗∗p < .005; ∗∗∗p < .0005.aOnly a preliminary draft of the report has been found with the help of another researcher.bWhen samples are small (e.g., fewer than 20, it can be of importance to use Hedges’ g as a correctionof Cohen’s d when calculating the effect of size (Hedges, 1981; Lipsey & Wilson, 2001).

The studies covered countries including Lebanon, France, Chile andthe United Kingdom. The children had been adopted at varied ages, frombefore 6 months to 4 years and later, in rare cases. The age, when tested,varied between 6 and 16 years, and four different measures of IQ were

Dow

nloa

ded

by [

Uni

vers

ity o

f W

ashi

ngto

n L

ibra

ries

] at

01:

22 1

9 N

ovem

ber

2014

Page 9: A Study of Adopted Children, Their Environment, and Development: A Systematic Review

Adopted Children, Their Environment, and Development 227

FIGURE 1 IQ for adopted adolescents and adolescents in care arrangements.

used. The studies were published over a 20-year period: 1973 to 1992. Thesedifferences may have caused a heterogeneity problem. The analysis showed asignificant evidence of heterogeneity (i.e., the statistical test of heterogeneityis statistically significant (p = .03) and I2 is estimated to 68%).

Instead of a fixed effects model, a random effects model assumes thatthe observed effect size differs from population mean by subject-level sam-pling error plus a random distributed error that represents other sources ofvariability associated with variation in procedures, settings, or measurementof IQ.

However, we find it safe to conclude (p < .00001) that the adoptedchildren showed significantly better cognitive development than their con-temporaries in the control groups who remained in a care environment.

School Achievement

Seven studies met the inclusion criteria for exploring the school achieve-ments of adopted children. The studies compared the developmental gainsfor adopted children with children who stayed in care (e.g., in a group carearrangement, an institution, with a foster family, or with a sibling). Threestudies were lacking information about standard deviation. Additional infor-mation was sought from the authors, but with no success. All three studiesgave significant results if we assume that the standard deviation of the mea-sures would have been on the same level as the other studies.

Figure 2 shows the results when these studies were excluded from themeta-analysis. Two out of the four studies showed an individual significantresult in favor of the adopted. The other two showed insignificant differencesbetween adopted and controls. Odds ratios measure the relationship for oddsfor positive school achievements for adopted children in relation to odds forpositive school achievements for controls. The odds ratio for the pooledstudies was estimated to be 0.54 (CI, 0.38–0.77) in favor of the adopted. Thecontrols had a 1.85 times higher risk of, for example, being in a special classor repeating a year or having learning disabilities (Table 2).

Dow

nloa

ded

by [

Uni

vers

ity o

f W

ashi

ngto

n L

ibra

ries

] at

01:

22 1

9 N

ovem

ber

2014

Page 10: A Study of Adopted Children, Their Environment, and Development: A Systematic Review

228 M. N. Christoffersen

TABLE 2 School Achievements for Adopted vs. Control Children

Country(N )

Age WhenAdopted

Age WhenTested

Test MeasureAverage/

PercentagesOddsRatio

(Bharat, 1997) India Youngerthan 1year

8–11 years SchoolAchievements

Adopted (41) 15.02Institution (24) 12.58(Palacios &

Sanchez-Sandoval,2005)

Spain Varied Teenagers School Behavior(CBI)

Adopted (273)Institution (198) 2.48(Maughan,

Collishaw, &Pickles, 1998)

England 2–4 years 11 years Achievements Testz-Scores

Adopted, Boys (57) 0.13Biological Parents,

Boys(101) −0.39 2.57

Adopted, Girls (33) 0.48Biological Parents,

Girls(122) −0.29 4.04

(Schiff, 1978) France Youngerthan 6months

6–14 years Special Class orRepeat a Year

Adopted (32) 13%Remained Siblings (20) 56% 8.52(Dumaret & Stewart,

1985)France Younger

than 6months

9–11 years Repeat 1 Year orMore

Adopted (35) 5.7%Remained Siblings (46) 37.6% 9.97Institution/Foster

Families(21) 85.7% 99.15

(Bohman, 1971) Sweden Before 1year

11 years Special Class

Adopted, Boys (90) 16%Adopted, Girls (73) 30%Biological Parents,

Boys(107) 9% 0.52

Biological Parents,Girls

(98) 12% 0.32

Foster Parents, Boys (69) 35% 2.83Foster Parents, Girls (55) 50% 2.33(Selwyn, 2006) England Learning

DisabilityAdopted (62–64) 39%Foster Parents (27) 59% 2.25

Note. CBI = Classroom Behavior Inventory.

Dow

nloa

ded

by [

Uni

vers

ity o

f W

ashi

ngto

n L

ibra

ries

] at

01:

22 1

9 N

ovem

ber

2014

Page 11: A Study of Adopted Children, Their Environment, and Development: A Systematic Review

Adopted Children, Their Environment, and Development 229

FIGURE 2 School achievements for adopted and remained adolescents.Note. Heterogeneity is evaluated in three ways: (1) with I 2, (2) with the Chi-square test ofheterogeneity, or (3) by comparing results of fixed and random effects models (Higgins,2002). Possible sources of heterogeneity were investigated, because there was evidence ofheterogeneity (i.e., the statistical test of heterogeneity is statistically significant and I 2 > 50%(Deeks et al., 2006). Both fixed effect and random effects models were examined.

The studies covered countries including France, Sweden, and the UnitedKingdom. The children had been adopted at younger than 1 year old. Onestudy gave no details about the adoption age. The age when tested variedbetween 6 and 14 years and two different measures of school achievementswere in use. The studies were published over a period covering more than30 years: 1971 to 2006. These differences may have caused a heterogeneityproblem. The analysis showed significant evidence of heterogeneity (i.e., thestatistical test of heterogeneity is statistically significant and I2 is estimatedto 86%). The heterogeneity could have been caused by differences betweentest measures or differences in the access to adoption or control groups.The studies did not allow us to make a meta-analysis to test whether thesedifferences between studies could explain the heterogeneity.

However, we find it safe to conclude (p < .0007) that the adoptedchildren showed a significantly better school achievement than their con-temporaries in the control groups who remained in a care environment,since none of the studies contradicted this conclusion.

Behavioral Disorder and Mental Health Problems

Six studies met the inclusion criteria for exploring the behavioral disordersand mental health problems of adopted children. The studies comparedthe developmental gains for adopted children with those who stayed behind(e.g., in a group care arrangement, in an institution, with a foster family, witha sibling, or with a single mother). Three studies gave significant results, oneof these in favor of the control groups.

Figure 3 and Table 3 show the results when these studies were ex-cluded from the meta-analysis. The remaining four studies show an individual

Dow

nloa

ded

by [

Uni

vers

ity o

f W

ashi

ngto

n L

ibra

ries

] at

01:

22 1

9 N

ovem

ber

2014

Page 12: A Study of Adopted Children, Their Environment, and Development: A Systematic Review

230 M. N. Christoffersen

FIGURE 3 Behavioral disorders and mental health problems for adopted and controlchildren.Note. Heterogeneity is evaluated in three ways: (1) with I 2, (2) with the Chi-square test ofheterogeneity, or (3) by comparing results of fixed and random effects models (Higgins,2002). Possible sources of heterogeneity were investigated if there was evidence ofheterogeneity (i.e., the statistical test of heterogeneity is statistically significant and I 2 > 50%;Deeks et al., 2006). Both fixed effect and random effects models were examined.

significant result in favor of the adopted. Odds ratios measure the relationshipfor odds for positive school achievements for adopted children in relationto odds for positive school achievements for the controls. The odds ratiowas estimated to be 0.61 (CI, 0.41–0.91) in favor of the adopted children.The controls had a 1.64 higher risk of exposing behavioral or mental healthproblems than the adopted.

The studies covered countries including Canada, New Zealand, Sweden,Denmark, and the United Kingdom. The children had been adopted atyounger than 1 year old in two studies; from 2 to 4 years in two studies; agevaried in two studies; and in one study no details about adoption age weregiven. The age when tested varied between 14 and 16 years in four studies,and in two studies adoptees were measured in their 20s or later. Six differentmeasures of behavioral problems and mental health problems were used.The studies were published over a 24-year period: 1971 to 1995. Thesedifferences may have caused a heterogeneity problem. The analysis showedsignificant evidence of heterogeneity (i.e., the statistical test of heterogeneityis statistically significant and I2 is estimated to 83%). However, we find it safeto conclude that the adopted children showed significantly fewer problematicbehavioral disorders or mental health problems than their contemporariesin the control groups who remained in institutional care (p = .01).

Self-Esteem

Six studies met the inclusion criteria for exploring the self-esteem of adoptivechildren. The studies compared the developmental gains of adopted children

Dow

nloa

ded

by [

Uni

vers

ity o

f W

ashi

ngto

n L

ibra

ries

] at

01:

22 1

9 N

ovem

ber

2014

Page 13: A Study of Adopted Children, Their Environment, and Development: A Systematic Review

Adopted Children, Their Environment, and Development 231

TABLE 3 Behavioral Problems and Mental Health Problems for Adopted vs. Control Children

Country (N )Age WhenAdopted

Age WhenTested

Test TypeResults in

Percentages Odds Ratio

(Bagley, 1991) Canada Average2.4 years

14 to15 years

Rutter’sDisorderScale

Adopted (37) 43.2%Biological Parents (23) 21.7% 0.36(Fergusson,

Lynskey, &Horwood, 1995)

NewZealand

14 to16 years

BehavioralProblems

Adopted (32) 21.9%Single Parents (60) 36.7% 2.07(Bohman &

Sigvardsson,1980)

Sweden Before 1year

15 years BehavioralProblems

Adopted. Boys (89) 4.5%Biological Parents,

Boys(108) 13.9% 3.43

Foster Family, Boys (112) 12.2% 2.95Adopted, Girls (71) 1.4%Biological Parents,

Girls(105) 15.2% 12.62

Foster Family, Girls (92) 7.6% 5.79(Hodges & Tizard,

1989)England 2–4 years 16 years Rutter’s B’

ScaleAdopted (21) 38%Biological Parents (9) 78% 5.78(Goodwin et al.,

1974)Denmark Younger

than 6weeks

20 to 51years

Psychopathology

Adopted Sons (20) 70%Remained Brothers (30) 23% 0.13(Triseliotis &

Russell, 1984)England 2 to 10 years In 20s Emotional

ProblemsAdopted (44) 25%Institution (40) 50% 3.00

with those children who stayed in care (e.g., in a group care arrangement,an institution, or with a foster family). Two studies reported significant dif-ferences in favor of the adopted, but information about standard deviationswas missing (Table 4). One study gave significant results in favor of thecontrol groups.

Figure 4 and Table 4 show the results when studies with missing infor-mation were excluded from the meta-analysis. The remaining three studiesshowed no significant differences. Odds ratios measure the relationship ofodds for positive self-esteem in adopted children in relation to odds forpositive self-esteem in controls.

Dow

nloa

ded

by [

Uni

vers

ity o

f W

ashi

ngto

n L

ibra

ries

] at

01:

22 1

9 N

ovem

ber

2014

Page 14: A Study of Adopted Children, Their Environment, and Development: A Systematic Review

232 M. N. Christoffersen

TABLE 4 Self-Esteem for Adopted vs. Control Children

Country (N )Age WhenAdopted

Age WhenTested

Measure forSelf-Esteem

Percentage andMeans Odds Ratio

(Fergussonet al., 1995)

NewZealand

14–16 years SEI LowSelf-Esteem

Adopted (32) 12.1%Biological Parents (60) 21.5% 1.99(Bagley &

Young, 1979)aEngland > 3 years 6–8 years Miller’s Measure

Adopted (60)Institution or

Foster Family(24)

(Bagley, 1991) Canada Average2.4 years

14–15 years Coopersmith >Lower Quartile

Adopted (37) 64.9%Biological Parents (23) 26.1% 0.19(Bagley, 1991) Canada Average

2.4 years14–15 years Self-Inflicted

BehaviorAdopted (37) 40.5%Biological Parents (23) 8.7% 0.14(Bharat, 1997) India > 1 year 8–11 years Global Self-WorthAdopted (41) 6.26Institution (24) 5.70(Palacios &

Sanchez-Sandoval,2005)b

Spain Varying Teenager Harter/Rosenberg

Adopted (90) 3.12Institution (156) 2.91 2.07(Selwyn, 2006) England Low Self-EsteemAdopted (62–64) 42%Foster Family (27) 37% 0.81

aBagley and Young (1979) have not reported measures on self-esteem.bPalacios and Sanchez-Sandoval (2005) responded by mail about standard deviations: Adopted SD =0.518, children in institution SD = 0.522.

FIGURE 4 Self-esteem for adopted and control children.

Dow

nloa

ded

by [

Uni

vers

ity o

f W

ashi

ngto

n L

ibra

ries

] at

01:

22 1

9 N

ovem

ber

2014

Page 15: A Study of Adopted Children, Their Environment, and Development: A Systematic Review

Adopted Children, Their Environment, and Development 233

The studies covered countries including Canada, New Zealand, India,Spain, and the United Kingdom. The children had been adopted youngerthan the age of 1 in one study; 2.4 years in one study; older than a year oldin two studies; varied ages in one study; and another gave no details at allabout the adoption age. The age when tested varied between 8 and 19 yearsin five studies; in one study there was missing information about the agewhen tested. Six different measures of self-esteem were in use. The studieswere published over a 17-year period: 1979 to 2006. These differences mayhave caused a heterogeneity problem.

Three studies had a binary outcome data and were analyzed together.Two studies were lacking outcome data, and the researchers were ap-proached to obtain the missing information. The analysis showed signifi-cant evidence of heterogeneity (i.e., the statistical test of heterogeneity isstatistically significant and I2 is estimated at 74%).

A preliminary conclusion is that the adopted children showed no sig-nificant differences in their self-esteem compared to their contemporaries inthe control groups who remained in the care environment (p = .19), but themissing information may influence this conclusion.

The results show that adoption seems to be a clearly better optionto institutionalization regarding cognitive development, for example, IQ(Colombo, Delaparra, & Lopez, 1992; Dennis, 1973; Hodges & Tizard, 1989),emotional problems (Triseliotis & Russell, 1984), and self-esteem (Palacios& Sanchez-Sandoval, 2005). Studies conducted in the years 1973 to 2006produced the same outcome in OECD countries and in Chile, Lebanon, andIndia.

Only a few studies compared adoption with foster care. The resultsshowed that adoption seems to be a better option to foster care regardingschool achievement (Bohman, 1971; Selwyn, 2006) and behavioral problems(Bohman & Sigvardsson, 1985) but the self-esteem results were insignificant(Selwyn, 2006).

DISCUSSION

The review project was conducted to examine the developmental gains ofadopted children compared to those placed in orphanages, institutions, andfoster care or compared to a control group of children from the same biolog-ical mothers. The main research question was: What would have happenedif the adopted children had remained in their families or in out-of-homeplacements? The question was operationalized to the following: Have chil-dren benefited from adoption compared to those children who remained ininstitutional or foster care?

The results indicated that the adopted children scored higher on IQ,school performance, mental health, and lack of behavioral problems than

Dow

nloa

ded

by [

Uni

vers

ity o

f W

ashi

ngto

n L

ibra

ries

] at

01:

22 1

9 N

ovem

ber

2014

Page 16: A Study of Adopted Children, Their Environment, and Development: A Systematic Review

234 M. N. Christoffersen

their non-adopted siblings or peers who stayed in institutions or with fosterparents, etc.

The results are more precise than former systematic reviews which com-pared adopted children with children in the new environment instead ofwith the children left in the environment from which the adopted childrenhad been removed. Since random access to adoption data is unethical andimpossible, the chosen method is the best available to answer the researchquestion. Still, the design of the present systematic review has some method-ological limitations.

First, there may be a positive or negative selection bias for adoption thatmay reflect in some studies that the most damaged children are selected foradoption or in other studies that the most damaged children are left in care.

Only one study had solved the selection problem by following all thechildren from an orphanage in Lebanon. In 1956, legislation allowed foradoptions and all the children were adopted from orphanages (Dennis,1973). The adopted were compared to the foundlings in the orphanageswho remained in institutional care. The early-adopted children had a normalcognitive development (IQ) and the late-adopted less than normal, while thefoundlings who remained in the institutions suffered serious loss in cognitivedevelopment.

Second, the Lebanon study disclosed another methodological problemin the present review. On the one hand, the gains of adoptions dependon the treatment of the controls. The developmental outcome of adoptionshas to be a relative measure; if control groups are exposed to deprivationand maltreatment, the gain of adoption will appear to be relatively positive.On the other hand, this method has the advantage of referring to realisticcontrols, which are the contemporary most relevant alternative to adoptions.

Third, evaluation of adoption based on old studies may overestimate thefuture gains because institutions and foster homes may be of much betterquality today.

Fourth, there is a problem with extrapolating experiences from countryto country because the treatment of controls may show huge variations. Weexpected to find considerable differences between countries and betweendifferent decades, but instead we found robust results. The results fromOECD countries such as Canada, Denmark, the United Kingdom, France,New Zealand, Spain, Sweden, and the United States were similar to the re-sults from studies in Chile, Lebanon, and India. And results from studies pub-lished in recent years were similar to those published two or three decadesago.

Finally, we may generalize a conclusion from the present study that thechildren seem to benefit from the attachment to their new parents and toovercome both the problematic upbringing before adoption and adapting toa new environment. If the adoption takes place early in life, the children seemto catch up and benefit from the new family. This generalization must be

Dow

nloa

ded

by [

Uni

vers

ity o

f W

ashi

ngto

n L

ibra

ries

] at

01:

22 1

9 N

ovem

ber

2014

Page 17: A Study of Adopted Children, Their Environment, and Development: A Systematic Review

Adopted Children, Their Environment, and Development 235

applied with some caution because some of results did not fully corroboratethis conclusion. We find it safe to conclude that the adopted children showeda significantly better cognitive development than their contemporaries whowere left in a care environment, but studies of the mental health problemsand self-esteem gave heterogeneous results.

One of the main problems in the reviewed studies was the heterogene-ity between studies exploring the same developmental outcome. Studies ofmental health problems showed significant heterogeneity. Two studies inparticular showed unexpected results (Goodwin et al., 1974; Bagley, 1991).When these studies were excluded from the aggregated test (Figure 3) theheterogeneity was reduced and the test for the overall effect improved.

The first study explored psychopathology in adopted and non-adoptedsons of alcoholics (Goodwin et al., 1974). The control group included broth-ers who stayed to be raised by alcoholic fathers. The design opened up forsome selection bias because children raised in severely disturbed alcoholicfamilies may have been placed out of the home and therefore excluded fromthe study.

The other study of adoption which gave divergent results was one ofNative children in Canada (Figure 4 and Table 4). It resulted unexpectedly inlower self-esteem among the adopted children compared to their siblings leftat home or in a care environment (Bagley, 1991). The researcher had madea similar study in England with opposite results (Bagley & Young, 1979).Bagley mentioned that the loss of self-esteem in the Canadian study couldhave been caused by conflicts over ethnicity. The adopted youngsters foundthemselves subjected to the stereotyping and rejection experienced by theaverage Native adolescent in urban Canada (Bagley, Young, & Scully, 1993).The interpretation of the study results emphasized that a positive outcomewith regard to mental health and self-esteem depends on the adopted chil-dren experiencing fewer role conflicts with regard to status, ethnic identity,and attachment to culture of origin.

We find it safe to conclude that the adopted children showed fewerproblematic behavioral disorders and mental health problems, but the con-clusion depends on the qualities of the new environment. When children areadopted into a hostile environment from an environment where cultural andfamily factors support their ethnic identity, the adoption can result in lowself-esteem and mental health problems compared to the siblings who havestayed behind. This conclusion needs testing in future work (Bagley et al.,1993). The studies in the present review did not include these potentiallydecisive environmental factors.

The present systematic review revealed the lack of systematic reviewsthat focus on developmental problems in adopted children compared to chil-dren left in the same hostile environment from which the adopted childrenwere removed. Future studies may examine and further clarify the protectiveinfluence of permanency that the adoptive placement provides.

Dow

nloa

ded

by [

Uni

vers

ity o

f W

ashi

ngto

n L

ibra

ries

] at

01:

22 1

9 N

ovem

ber

2014

Page 18: A Study of Adopted Children, Their Environment, and Development: A Systematic Review

236 M. N. Christoffersen

NOTES

1. An exception to this rule is the study by Wayne Dennis in which Tunisian children adopted inTunisia and in the United States were followed and their cognitive development was compared to thenon-adopted children in the Tunisian institution (Dennis, 1973).

2. A specified search strategy is published in Christoffersen et al. (2007, pp. 161–162).

REFERENCES

Bagley, C. (1991). Adoption of native children in Canada: A policy analysis and aresearch report. In H. Altstein & R. J. Simon (Eds.), Intercountry adoption: Amultinational perspective (pp. 55–79). New York, NY: Praeger Publishers.

Bagley, C., & Young, L. (1979). The identity, adjustment, and achievement of tran-sracially adopted children: A review and empirical report. In G. K. Verma &C. Bagley (Eds.), Race, education, and identity (pp. 192–219). London, UK:Macmillan.

Bagley, C., Young, L., & Scully, A. (1993). International and transracial adoptions.A mental health perspective. Aldershot, Hants, England: Avebury.

Bharat, S. (1997). Intellectual and psychosocial development of adopted children.Mumbai, India: Tata Institute of Social Scences.

Bimmel, N., Juffer, F., van IJzendoorn, M. H., & Bakermans-Kranenburg, M. J. (2003).Problem behavior of internationally adopted adolescents: A review and meta-analysis. Harvard Review of Psychiatry, 11, 64.

Bohman, M. (1971). A comparative study of adopted children, foster children andchildren in their biological environment born after undesired pregnancies. ActaPaediatrica Scandinavica, Supplement, 221, 1–38.

Bohman, M., & Sigvardsson, S. (1980). A prospective, longitudinal-study of childrenregistered for adoption: A 15-year follow-up. Acta Psychiatrica Scandinavica,61, 339–355.

Bohman, M., & Sigvardsson, S. (1985). A prospective longitudinal study of adop-tion. In Longitudinal studies in child psychology and psychiatry (pp. 137–155).Chichester, UK: Wiley.

Chinn, S. (2000). A simple method for converting an odds ratio to effect size for usein meta-analysis. Statistics in Medicine, 19, 3127–3131.

Christoffersen, M. N., Hammen, I., Andersen, K. R., & Jeldtoft, N. (2007). Adoptionsom indsats: en systematisk gennemgang af udenlandske erfaringer. Copen-hagen, Denmark: SFI–Det Nationale Forskningscenter for Velfærd.

Cohen, N. J. (2002). Adoption. In M. Rutter & E. Taylor (Eds.), Child and adolescentpsychiatry (4th ed., pp. 373–381). Oxford, UK: Blackwell.

Colombo, M., Delaparra, A., & Lopez, I. (1992). Intellectual and physical outcomeof children undernourished in early life is influenced by later environmental-conditions. Developmental Medicine and Child Neurology, 34, 611–622.

Deeks, J. J., Higgins, J. P. T., & Altman, D. (2006). Analysing and presenting results.In J. P. T. Higgins & S. Green (Eds.), Cochrane handbook for systematic reviewsof interventions. Chichester, UK: John Wiley & Sons.

Dow

nloa

ded

by [

Uni

vers

ity o

f W

ashi

ngto

n L

ibra

ries

] at

01:

22 1

9 N

ovem

ber

2014

Page 19: A Study of Adopted Children, Their Environment, and Development: A Systematic Review

Adopted Children, Their Environment, and Development 237

Dennis, W. (1973). Children of the creche. New York, NY: Appleton-Century-Crofts.Dumaret, A., & Stewart, J. (1985). IQ, scholastic performance and behavior of sibs

raised in contrasting environments. Journal of Child Psychology and Psychiatryand Allied Disciplines, 26, 553–580.

Fergusson, D. M., Lynskey, M., & Horwood, L. J. (1995). The adolescent outcomesof adoption: A 16-year longitudinal study. Journal of Child Psychology andPsychiatry and Allied Disciplines, 36, 597–615.

Goodwin, D. W., Schulsin, F., Moller, N., Hermanse, L., Winokur, G., & Guze, S.B. (1974). Drinking problems in adopted and nonadopted sons of alcoholics.Archives of General Psychiatry, 31, 164–169.

Hedges, L. V. (1981). Distribution theory for Glass’s estimator of effect size andrelated estimators. Journal of Educational Statistics, 6, 107–128.

Higgins, J. P. T. (2002). Quantifying heterogeneity in a meta-analysis. Statistics inMedicine, 21, 1539–1558.

Hodges, J., & Tizard, B. (1989). IQ and behavioural adjustment of ex-institutionaladolescents. Journal of Child Psychology and Psychiatry, 30, 53–75.

Juffer, F., & van IJzendoorn, M. H. (2005). Behavior problems and mental health re-ferrals of international adoptees: A meta-analysis. JAMA: Journal of the AmericanMedical Association, 293, 2501–2515.

Juffer, F., & van IJzendoorn, M. H. (2007). Adoptees do not lack self-esteem: A meta-analysis of studies on self-esteem of transracial, international, and domesticadoptees. Psychological Bulletin, 133, 1067–1083.

Lipsey, M. W. & Wilson, D. B. (2001). Practical meta-analysis. Thousand Oaks, CA:Sage Publications.

Maughan, B., Collishaw, S., & Pickles, A. (1998). School achievement and adultqualifications among adoptees: A longitudinal study. Journal of Child Psychologyand Psychiatry, 39, 669–685.

Palacios, J., & Sanchez-Sandoval, Y. (2005). Beyond adopted/nonadopted compar-isons. In D. M. Brodzinsky & J. Palacios (Eds.), Psychological issues in adoption(pp. 117–144). Westport, CT: Praeger Publishers.

RevMan. (2008). Review Manager [Computer program]. (Version 5.0.) [Computersoftware]. Copenhagen, Denmark: The Nordic Cochrane Centre, The CochraneCollaboration.

Schiff, M. (1978). Intellectual status of working-class children adopted early intoupper-middle-class families. Science, 200, 1503–1504.

Selwyn, J. (2006). Costs and outcomes of non-infant adoptions. London, UK: BAF.Triseliotis, J., & Russell, J. (1984). Hard to place. The outcome of adoption and

residential care. London, UK: Heinemann Educational Books.van IJzendoorn, M. H., Juffer, F., & Poelhuis, C. W. K. (2005). Adoption and cog-

nitive development: A meta-analytic comparison of adopted and nonadoptedchildren’s IQ and school performance. Psychological Bulletin, 131, 301–316.

van IJzendoorn, M. H., Bakermans-Kranenburg, M. J., & Juffer, F. (2007). Plasticityof growth in height, weight, and head circumference: Meta-analytic evidenceof massive catch-up after international adoption. Journal of Developmental &Behavioral Pediatrics, 28, 334–343.

Dow

nloa

ded

by [

Uni

vers

ity o

f W

ashi

ngto

n L

ibra

ries

] at

01:

22 1

9 N

ovem

ber

2014