The influence of perinatal complications and environmental adversity on boys' antisocial behavior

12 Pages·2004·0.13 MB·English

JournalofChildPsychologyandPsychiatry46:1(2005),pp35–46 doi:10.1111/j.1469-7610.2004.00336.x
The influence of perinatal complications
and environmental adversity
on boys’ antisocial behavior
Joy E. Beck and Daniel S. Shaw
UniversityofPittsburgh,USA
Background: The purpose of the present study was to test components of Raine’s (2002) biosocial
model, specifically the interactive effects of perinatal complications, rejecting parenting, and family
adversityonthedevelopmentofearly-onsetantisocialbehavior(ASB).Boys’internalizingproblemswere
alsotestedtoinvestigatethespecificityofthemodel. Methods: Birthrecordsinadditiontolongitudinal
data were collected on 310 low-income boys followed from birth until 10years of age.
Results: Findings demonstrated support for a biosocial framework in predicting ASB but not inter-
nalizing problems. Family adversity, and to some extent rejecting parenting, consistently predicted
youthoutcome.PerinatalcomplicationsemergedasapredictorofASBbutonlyinthecontextofother
family risk factors. According to maternal report, boys experiencing high levels of perinatal complica-
tions,rejectingparenting,andfamilyadversityshowedsignificantlyhigherlevelsofASBthanboyswith
lowerlevelsoftheseriskfactors.Thisfindingwaspartiallycorroboratedbyyouthself-report,suchthat
boys experiencing high levels of perinatal complications and family adversity reported more antisocial
activitythanboysexperiencingnoriskorriskinonlyonedomain. Conclusions: Thisstudysupports
the specific prediction of ASB in middle childhood from a biosocial model. Findings also highlight the
salience of a negative psychosocial environment on childhood maladjustment. Intervention efforts
includingparentingskillsandcopingstrategiesformothersofchildrenfrommultipleriskenvironments
are advocated. Keywords: Perinatal complications, antisocial behavior, family adversity, high-risk
sample. Abbreviations: ASB: antisocialbehavior.
Recently there has been an increased interest in constructs for increasing the prediction of adult
antisocial behavior that emerges early in childhood antisocial behavior. However, from a developmental
and persists into adulthood. Contributing to this perspective, two important linkages in the model
interest is the growing evidence that 5–6% of the have received modest or only superficial attention
most persistent offenders are responsible for 50% with respect to childhood-onset ASB. First, few
of known crimes (Aguilar, Sroufe, Egeland, & studies have examined whether the biosocial model
Carlson, 2000). Moreover, individuals performing predictsbehaviorproblemsinmiddlechildhood.For
antisocial acts are difficult to rehabilitate and the two studies that have investigated childhood
are likely to become recidivistic (Kazdin, 1987; outcomes of ASB, the results for an interaction be-
Moffitt, 1993a). Thus, identifying risk factors of tween perinatal complications and family adversity
early antisocial behavior has important implica- have proven inconsistent (Arseneault et al., 2002;
tions for enhancing both intervention and preven- Laucht et al., 2000). Second, the measures of envir-
tion efforts. onmental risk used in these studies typically have
Raine (2002) has proposed a biosocial model to been distal. In this vein, the main goal of the pres-
account for the contribution of both biologically and ent study was to test components of the biosocial
environmentally related risk factors on the devel- model on early-onset ASB. It was hypothesized that
opment of antisocial behavior (ASB). One group of the association between perinatal complications
studies testing the biosocial model has focused on and child ASB would be moderated by high levels of
perinatal complications and environmental adver- family risk factors, such that the adverse effects of
sity, noting a consistent interaction between the perinatal complications on boys’ ASB would be
presence of both of these factors and the develop- evident only under conditions of high family adver-
ment of serious antisocial behavior in adulthood sity. The following sections review literature per-
(Arseneault, Tremblay, Boulerice, & Saucier, 2002; taining to perinatal complications and constructs of
Piquero & Tibbetts, 1999; Raine, Brennan, Med- environmental adversity used in the present study,
nick, & Mednick, 1997). Research investigating namely rejecting parenting and cumulative family
the biosocial model has illuminated potential adversity.
(cid:1)AssociationforChildPsychologyandPsychiatry,2004.
PublishedbyBlackwellPublishing,9600GarsingtonRoad,OxfordOX42DQ,UKand350MainStreet,Malden,MA02148,USA
36 JoyE.BeckandDanielS.Shaw
developmentofearly-onsetASB(Campbell,Shaw,&
Perinatal complications
Gilliom, 2000; Shaw, Ingoldsby, Gilliom, & Nagin,
Research has suggested that there are multiple risk 2003). Parental responsiveness, sensitivity to social
factors and pathways associated with the develop- cues, and emotional availability have all been as-
ment of antisocial behavior during early and middle sociated with positive outcomes in young children,
childhood (Cicchetti & Rogosch, 1996). One such such as behavioral regulation and social compet-
risk factor is maternal health status which, when ence (Bost, Vaughn, Washington, Cielinski, &
compromised during pregnancy, has been asso- Bradbard, 1998; Martin, 1981; Wakschlag & Hans,
ciated with impaired functioning of offspring’s cen- 1999). A lack of parental responsiveness during
tralnervoussystem(CNS)andsubsequentproblems infancy, however, has been associatedwith negative
in the child’s well-being (Moffitt, 1993a, 1993b). outcomes, such as ASB, later in childhood (e.g.,
Complicationsduringtheprenatal(conceptionto7th Shaw, Keenan, & Vondra, 1994a; Shaw et al.,
month of pregnancy) and perinatal (7th month of 1998b; Wakschlag & Hans, 1999). Parental rejec-
pregnancy through 28 days after birth) periods are tion, the combination of harsh and controlling
early factors affecting CNS development and have parenting practices coupled with unacceptance of
been tested individually as predictors of deviant the child, also has been linked with the develop-
outcomes (for a review see Brennan & Mednick, ment of later ASB (Campbell, Pierce, Moore, &
1997). The most consistent relations have been Marakovitz, 1996; Dishion, 1990; Dodge, Pettit, &
found between complications during the perinatal Bates, 1994; Younge, Oetting, & Deffenbacher,
stage and later ASB (Kandel & Mednick, 1991). Di- 1996). The present study investigated a blend of
rect relations between perinatal complications and these two parenting factors, which from this point
ASB have not typically been demonstrated (Cohen, will be referred to as ‘rejecting parenting’ for the
Velez, Brook, & Smith, 1989; Rantakillio, Koiranen, sake of brevity.
& Moettoenen, 1992). However, in the context of
family adversity, high levels of perinatal complica-
tions have been associated with increased risk of Cumulative family adversity. A number of studies
child ASB (Broman, Nichols, & Kennedy, 1975; support a ‘cumulative risk hypothesis,’ wherein the
Drillien, 1964; Werner, Bierman, & French, 1971). number of environmental stressors rather than the
Recentempirical research testingthe biosocialin- particular combination of stressors, has been asso-
teractionhypothesishasdemonstratedthatrelations ciated with child behavior problems both concur-
between perinatal complications and later ASB are rently and longitudinally (Deater-Deckard, Dodge,
moderated by environmental adversity (Arseneault Bates,&Pettit,1998;Rutter,Cox,Tupling,Berger,&
et al.,2002;Lauchtet al.,2000;Piquero&Tibbetts, Yule, 1975a; Rutter et al., 1975b; Sameroff, Seifer,
1999; Raine, Brennan, & Mednick, 1994, 1997; Zax,&Barocas,1987;Sanson,Oberklaid,Pedlow,&
Werner, 1987). In a Danish birth cohort of males, Prior,1991;Shaw,Vondra,Hommerding,Keenan,&
Raine and colleagues (1994, 1997) found that boys Dunn, 1994b; Shaw, Winslow, Owens, & Hood,
whosuffered bothperinatal complicationsandearly 1998a).Intheirclassicstudy,Rutterandcolleagues
maternalrejectionweremostlikelytobecomeviolent (1975a, 1975b) found a dramatic rise in the prob-
offenders in adulthood. Arseneault and colleagues abilityofchildadjustmentdifficultiesasthenumber
(2002)alsofoundsupportforthebiosocialmodelina of family stressors increased. Sameroff and his col-
low-income sample of 849 boys. Their results dem- leagues (1987) tested the impact of three sets of
onstrated that a combination of perinatal complica- variables on children’s behavior and found that
tions posing imminent harm to the infant predicted children with high multiple environmental risk
elevated rates of physical aggression at ages 6 and scoreshadmuchworseoutcomesthanchildrenwith
17 years in boys when children were reared in lowmultipleriskscores.Intwolongitudinalstudies,
impoverishedenvironments. Shaw and colleagues (1994b, 1998a) also demon-
strated support for the ‘family adversity hypothesis’
in early childhood by examining its effects on boys
Environmental risk
starting in infancy. They showed that boys’ inter-
A substantial amount of research links aspects of a nalizingandexternalizingbehaviorproblemsatages
child’s environment to the development of ASB. 2to3.5increasedinalinearfashionwithrespectto
Both proximal and distal factors of the child’s the number of family stressors present in the first
environment have been related to ASB, including two years.
parenting and the accumulation of less proximal The goal of the present study was to test com-
environmental risk factors (e.g., low SES and ponents of the biosocial model proposed by Raine
marital conflict). (2002), specifically examining the interactive effects
of perinatal complications, rejecting parenting, and
Parenting. A number of studies suggest that the family adversity on the development of early-onset
quality of early parental care, such as unrespons- ASB. This research has the potential to advance
iveness and rejection, plays a significant role in the the biosocial model because of the examination of
Theinfluenceofperinatalcomplications 37
early environmental risk factors, the use of mul- Methods
tiple informants and repeated measures, and a test
Participants
of specificity. Prospective data were collected from
birth to 10 years of age on a racially diverse, high- Families were selected from the Pitt Mother and Child
risk sample of boys from a metropolitan area in the Project, an ongoing longitudinal study aimed at exam-
United States. Unlike previous studies which have ining the antecedents of children’s antisocial behavior.
used distal proxies of maternal rejection, such as Initially, 310 families with boys between the ages of 6
and17 monthswererecruitedfromWomen,Infant,and
an attempt to abort the fetus (Raine et al., 1997),
Children(WIC)NutritionalSupplementProgramclinics
the present study measured maternal parenting
in the metropolitan Pittsburgh area. The WIC program
using observational and interview data when the
provides nutritional resources to income-eligible famil-
boys were 2 years old. In previous studies, meas-
ies. In addition to having a son between the ages of 6
ures of environmental adversity have been based
and 17months, the families were required to have an-
on indices of socioeconomic status (e.g., Arseneault other sibling living at home. Mothers and their sons
et al., 2002). Our study improved upon this prac- wereformallyassessedforthefirsttimewhenthetarget
tice by creating a cumulative family adversity index child was 1.5 years old. The sample consists of 40%
comprised of several proximal facets of the child’s African-American, 54% Caucasian families, and 6%
home environment measured prospectively and other. At recruitment, the average educational level of
repeatedly over time. Child internalizing problems mothers was 12.5 years and mean per capita income
was $2,892 per year. Observations of parent–child
were included to explore the specificity of the bio-
interaction and maternal- and youth-report data were
social model to child ASB (Laucht et al., 2000).
collected at the lab and/or home when the target child
Specifically, we hypothesized that data on proximal
was1.5,2,3.5,8,and10yearsold(i.e.,youthreportat
family processes (parenting and quality of home
age 10only).
environment) would increase our understanding of
All boys whose birth records were obtained and
potential mechanisms underlying the interaction whose families participated in the age 1.5-, 2-, and/or
between perinatal complications and family advers- 3.5-year visits, as well as the age 8- and/or 10-year
ity, an issue that to date has only examined the visit,wereselectedforinclusioninthisstudy.Basedon
quality of the family environment from a distal thesecriteria,approximately250families,or81%ofall
perspective (e.g., socioeconomic status of family families at recruitment, were included in the analyses.
versus having observations of parenting behavior). Comparisonsofboyswhosebirthrecordswereobtained
versus those whose were not available revealed no sig-
Three hypotheses were tested in relation to later
nificantdifferencesonanyindexofchildadjustmentat
child ASB. First, based on the evidence reviewed
ages8 and10 years.
above (e.g., Kandel & Mednick, 1991; Raine et al.,
1994, 1997), perinatal risk was not expected to be
Procedures
directly related to later ASB. However, based on
previous research (e.g., Sameroff et al., 1987; Shaw During the age 1.5- and 2-year laboratory visits,
et al., 1994a, 1998b), direct relations were anticip- mothers completed questionnaires, provided demo-
ated to emerge between rejecting parenting, family graphic information, and engaged in interactive tasks
adversity, and ASB. Second, in accordance with with their sons. A home assessment at age 2, which
biosocial theory and previous studies (e.g., Arsen- tookplaceonthesamedayasthelaboratoryvisit,was
conducted to gather observational data on the home
eault et al., 2002; Raine et al., 1994, 1997), we
environment and parenting. The age-3.5 assessment
predicted that children with risk factors from both
consisted of one laboratory visit, in which mothers
biological and environmental domains would be at
completed questionnaires and participated in several
greater risk for later ASB. Specifically, children
structured tasks with their child. At the age 8 and 10
who experienced high levels of perinatal complica-
home assessments, mothers and sons completed
tions in addition to high levels of rejecting parent- questionnaires about the boys’ behavior with the
ing or family adversity were expected to be at examiner.Alllaboratoryvisitswerevideotapedthrough
greater risk for later ASB. Third, based on previous aone-way mirror.
research (Rutter et al., 1975a, 1975b; Sameroff
et al., 1987; Shaw et al., 1994b, 1998a), we Measures
anticipated the effects of perinatal complications
Predictor variables
on child ASB to be most evident within the con-
Perinatal complications. Written consent to access
text of maximum family adversity. In particular,
birthrecordswasobtainedfrommothersatthetimeof
we expected a significant three-way interac-
the first lab assessment when the boys were 1.5years
tion, such that children who experienced high
old, and admission to medical records was requested
levels of perinatal complications along with high
directly from hospitals. The first author and research
levels of rejecting parenting and family adver- assistants visited hospitals in the metropolitan area
sity would demonstrate higher levels of later ASB and copied pertinent information from the mothers’
than children with only one or two of these risk and/or boys’ medical records. Some records were
factors. obtained from hospitals via postal mail. Seventy-six
38 JoyE.BeckandDanielS.Shaw
percent of birth records were obtained from nine information about the EPCS, see Shaw and col-
teaching/university-based hospitals. The remaining leagues, 1998b). The abridged 6-item HOME Respon-
24%wereobtained fromsix community hospitals. sivity factor and the EPCS rejecting parenting score
Thepresentstudyemployedthesameweightedscale were negatively correlated (r ¼).26, p<.001). To
system for perinatal complications used in previous create the single measure of ‘rejecting parenting’ from
research studies (Kandel & Mednick, 1991; Mednick, the constructs of maternal unresponsiveness and re-
Hocevar, Baker, & Teasdale, 1983; Raine et al., 1996, jection, HOME Responsivity standardized scores were
1997).1 This weighted-severity scale was developed by subtracted from EPCS rejecting parenting standard-
thecollaborationofAmericanandDanishobstetricians ized scores.
and pediatric neurologists and was derived from a
‘logical and clinical approach’ as opposed to factor or
cluster analysis techniques. The complete list of com-
Cumulative family adversity index. The family
plications and the severity ratings for this weighted
adversityriskindexwascomprisedofsevenconstructs:
scoringsystemwereobtainedfromtheauthorsandare
family size, family income, family criminality, neigh-
presented in the Appendix. Items were weighted on
borhooddangerousness,numberofstressfullifeevents,
a scale of 0–5 according to the judged severity of
parent daily hassles, and parental conflict. These fac-
the complication or deviance, with the most severe
tors were chosen for their potential effects above and
complication score beingrecorded.
beyond those accounted for by parenting. Information
pertaining to family income, parental criminality, and
Home Observation for Measurement of the Envir-
overcrowding was obtained from a demographic ques-
onment Inventory (HOME; Caldwell & Bradley, tionnaire that was administered at recruitment and at
1984) and Early Parenting Coding System (EPCS; the age 2- and 3.5-year assessments. The number of
Winslow&Shaw,1995). The infant/toddler version negativelifeevents(e.g.,lossofjob)occurringinthelast
of the HOME was administered as part of the age-2 yearwasmeasuredbytheNegativeLifeEventsfactorof
home assessment. The HOME is a 45-item instrument theLifeExperiencesSurvey(LES;Sarason,Johnson,&
that assesses the quality and quantity of support and Siegel, 1978), administered at the age 2 visit. The Par-
stimulation in the child’s home environment through entingDailyHassles(PDH;Crnic&Greenberg,1990)is
observational and parent interview components and a 20-item measure of typical everyday events parents
has demonstrated good reliability and validity (Cald- encounterwithchildren.Thefrequencyandintensityof
well & Bradley, 1984). Six items tapping parental the items were summed and used in the analyses.
unresponsiveness were selected from the Emotional Mothers completed at the PDH at the age-1.5, -2, and
and Verbal Responsivity subscale and subjected to -3.5 visits, and scores were averaged across the avail-
factor analysis, with factor loadings ranging from .55 able (potentially 3) time points. The Neighborhood
to .74 (a ¼.69). Lower scores indicate less respon- Questionnaire (NQ; Pittsburgh Youth Study, 1991) is a
sivity. The EPCS was designed to measure a range of 17-itemmeasureofproblematicanddangerousactivit-
parenting behaviors typically exhibited in interactions ies within a family’s neighborhood as perceived by the
with young children. Molecular and global ratings parentattheage-2visit.Theitemsaresummedtocreate
were recorded from videotaped mother–child interac- one factor score for neighborhood dangerousness. Par-
tions during a structured clean-up task at the age-2 ental conflict was measured using the verbal and
laboratory assessment that took place the same day physicalaggressionfactorsoftheConflictTacticsScales
as the HOME assessment. A composite of two (CTS-FormN;Straus,1979),administeredattheage3.5
molecular (j¼ .79–.87) and three global (j¼ .83–.94) visit. There was adequate reliability for the LES, PDH,
ratings relevant to rejecting parenting was employed. andCTS(a ¼.63to.90).
Higher scores indicate higher rejection (for more To create a cumulative Family Adversity Index (FAI)
score, each of the seven measures listed above were
assignedascoreofone(1)iftheymetthefollowingcri-
1There have been numerous ways in which perinatalcompli- teria (indicating greater adversity), and zero (0) if they
cationshavebeengroupedandusedbyresearchers.Raineand didnot.Thecriterionforlowincomewasanannualpro-
colleagues(1994,1997)usedthesamesystemasthepresent rated per capita income equal to or less than $2,500,
study but dichotomized the variable into zero versus one or averaged over three time points. Overcrowding was de-
more birth complications. Arseneault and colleagues (2002) fined as families having four or more children living at
ran a principal components analysis to derive three factors. home or having less than one room per person at any
Although both groups of researchers used different systems,
time point (Rutter et al., 1975b; Shaw etal., 1998a).
they found a significant biosocial interaction predicting to
Parental criminality was based on maternal report of
antisocialbehavior.Wesubjectedouritemstoafactoranalysis
criminal behavior committed by the child’s family
andobtainedafactorsimilartoArseneaultetal.’sDeadlyRisk
members living in the home during the child’s lifetime.
Situation factor. Our weighted-severity perinatal score was
FortheLES,PDH,NQ,andCTS,thecriterionwassetat
significantly correlated with this factor (r¼.23, p<.001), as
orgreaterthanonestandarddeviationabovethesample
well as a weighted-frequency score (r¼.77, p<.001) and
mean.Thesevenscoresweresummed,withtotalscores
a continuous frequency-of-complications score (r¼.65,
ranging from0 to7. In thecase of missing datafrom a
p<.001; Richardson, Hamel, Goldschmidt, & Day, 1999).
Since one classification system has not proven consistently repeatedmeasure,datafromanyofthetimepointswas
superioroveranotherinpredictingantisocialbehavioranddue usedonitsown.Thetotalsummedscorewasdividedby
tothehighcorrelationofthepresentsystemwiththesystems the total number of measures obtained, since some
employed in other studies, we preferred to have continuous subjects were missing data from a questionnaire
variableemphasizingpotentialseverityofeachcomplication. administeredatonlyonetimepoint(e.g.,LESorCTS).
Theinfluenceofperinatalcomplications 39
Child outcome variables. Child Behavior Checklist– Results
ParentReportForm(CBCL;Achenbach,1991). TheCBCL
For all analyses, the CBCL Externalizing and Inter-
is a reliable and well-validated 112-item instrument
thatassesseschildbehaviorproblems.Parentsratethe nalizing broad-band factors at age 8 years and the
items on a three-point scale ranging from ‘Not True’ to SRD and child internalizing score at age 10 years
‘VeryTrue’oftheirchildinthepastsixmonths.Asboth served as the dependent variables.
aggressivereactiveandcovertdelinquentactivitieshave
been linked to perinatal complications and family risk
Descriptive statistics and intercorrelations
factors (Arseneault et al., 2002; Raine et al., 1994,
1997),thecurrentstudyusedtheExternalizingbroad-
Descriptive statistics for all study variables are pro-
band factor to measure later child ASB at age 8. The
vided in Table 1. The mean severity weight for peri-
Internalizing broad-band factor was also used to
natalcomplicationswas2.25(median ¼ 3).Thesum
investigate specificity effects when boys were 8years
ofindividualcomplications(i.e.,unweighted)ranged
old.
from 0 to 8, with the median being 2. The assigned
severity weight was significantly correlated with
Self-Report of Delinquency (SRD; Elliott, Huizinga,
the number of perinatal complications (r ¼ .74,
& Ageton, 1985). Ten items (a¼ .71) were adapted
p < .001). The mean t scores for maternal-rated
fromtheSRDtomeasurechildren’sreportofantisocial
behavior at age 10. The SRD is a semistructured Externalizing (M ¼ 51, SD ¼ 10.2) and Internalizing
interview that assesses the frequency with which an (M ¼ 49, SD ¼ 9.8) were comparable to the norma-
individualhasengagedindelinquentbehavior.Usinga tive mean t score of 50 (Achenbach, 1991).
3-point rating system (1 ¼‘never,’ 2 ¼‘once/twice,’ Two-tailed Pearson product intercorrelations were
3 ¼‘more often’), boys rated the extent to which they computed among the predictor variables: perinatal
engaged in different types of antisocial behaviors (e.g., complications, rejecting parenting, and the family
stealing,throwingrocksorbottlesatpeople)withinthe
adversity index (FAI) and among child outcome var-
lastyear.
iables across informants. Rejecting parenting and
FAI were modestly but significantly related (r ¼ .23,
Child Depression Inventory (CDI; Kovacs, 1992)
p < .001), but neither was related to perinatal com-
and Multidimensional Anxiety Scale for Children
plications.Maternalreportsofchildadjustmentwere
(MASC; March et al., 1997). Boys completed 10-
modestly related to youth-reports (r ¼ .15 for Exter-
item short forms of the CDI (a ¼.66) and MASC
nalizing/Delinquency, and r ¼ .17 for Internalizing,
(a ¼.69) at age 10. For the items on the CDI, boys
p < .05). Across informants, reports of Externaliz-
were presented with a group of three statements and
ing/Delinquency were not related to reports of
were asked to choose the sentence that best described
their feelings in the past two weeks. These items were Internalizing.
summed to generate one measure indicative of child
depressive symptomatology. For the items on the
Testing direct relations between perinatal
MASC,boyswerepresentedwithaseriesofstatements
complications, rejecting parenting,
indicating anxiety-arousing situations and were asked
to rate how true each statement was for him on a 4- family adversity, and later child adjustment
point scale. These items were summed to form one
To examine the hypothesis that perinatal complica-
factorindicativeofanxietysymptomsinchildren.Both
tionswouldnotbedirectlyrelatedtolaterantisocial
the CDI and MASC have been shown to have adequate
behavior (ASB) and that rejecting parenting and
reliability and validity (Kazdin, French, Unis, Esveld-
family adversity would be directly and positively
Dawson, & Sherick, 1983; March etal., 1997). The
total CDI and MASC scores were standardized and related to later ASB, a series of one-tailed Pearson
then averaged (r¼ .30, p< .001) to create a measure product correlation coefficients were calculated be-
of child internalizing problems. tween these factors and CBCL Externalizing at age
Table1 Descriptivestatisticsforpredictorsandoutcomemeasuresofboys’adjustment
Variables Mean Standarddeviation Range Samplesize
Predictors
Perinatalcomplications 2.25 1.32 0–5 271
Rejectingparenting .0026 1.34 )2.04–4 286
Familyadversityindex .2777 .203 0–1 307
Outcomemeasures
Maternalreport–age8
Externalizing(CBCL) 50.98 10.2 30–82 248
Internalizing(CBCL) 48.59 9.78 34–80 248
Youthreport–age10
Delinquency(SRD10-items) 1.74 2.29 0–12 234
Internalizing(CDI/MASCz-score) ).0004 ).144 )1.4–3.24 234
40 JoyE.BeckandDanielS.Shaw
8 yearsandtheSelf-ReportofDelinquencyatage10, factors on child outcomes. A series of hierarchical
the results of which are presented in Table 2. The regression analyses were computed to examine
correlations between the predictor variables and whether high levels of perinatal complications
Externalizing/Delinquency were generally consist- would interact with high levels of rejecting par-
ent with expectations. Perinatal complications were enting and/or family adversity to increase risk of
unrelated to Externalizing/Delinquency, and rela- later child adjustment problems. For all regression
tions between the FAI and child adjustment were analyses, independent variables were hierarchically
significant across informants (r ¼ .22 and .20 for entered into regression equations based on chro-
mothers and boys, respectively, p < .001), albeit nological order: perinatal complications followed by
modest in magnitude. Rejecting parenting at age 2 rejecting parenting, then family adversity, followed
was significantly related to child-reported De- by each of the three two-way interaction terms (i.e.,
linquency eight years later (r ¼ .17, p < .01). perinatal complications · rejecting parenting,
Totestforthespecificityoftheseeffectsinrelation followed by perinatal complications · family
to child adjustment, one-tailed Pearson product adversity, and then rejecting parenting · family
correlations were computed between predictor vari- adversity), and finally the three-way interaction
ables and maternal- and child-rated Internalizing at term.
ages8and10,respectively.AsdisplayedinTable 2,
perinatalcomplicationsandrejectingparentingwere Externalizing/Delinquency
consistently unrelated to Internalizing, whereas FAI Table 3 displays the summary of the hierarchical
wassignificantlycorrelatedwithbothmaternal-and regression analyses involving perinatal complica-
child-report Internalizing (r ¼ .21 and .12, p < .001 tions, rejecting parenting, and FAI for maternal-
and .05, respectively). rated Externalizing at age 8. A significant 3-way
interaction emerged in relation to Externalizing. To
explore the nature of this effect, post hoc analyses
Testing for interactive effects of the predictor
using one-way ANOVAs were conducted. Risk
variables on child adjustment
groups were generated by parsing the environmen-
The second issue of interest was testing the inter- tal risk variables into ‘high’ and ‘low’ scores using
active effects of biological and environmental risk the upper quartile asthe cutoff. That is, high scores
Table2 One-tailedPearsoncorrelationsbetweenpredictorvariablesandboys’adjustmentbymaternalandyouthreport
Outcome Perinatal Rejecting Familyadversity
measure complications parenting index
Maternalreport–age8
CBCLExternalizing Pearsonr ).004 .064 .218**
N 216 237 248
CBCLInternalizing Pearsonr ).016 .040 .211**
N 216 237 248
Youthreport–age10
SRDDelinquency Pearsonr .049 .174** .202**
N 207 223 234
CDI/MASCInternalizing Pearsonr ).087 .037 .124*
N 207 223 234
*p<.05;**p<.01.
Table3 Summaryofregressionanalysesforperinatalcomplicationsandenvironmentalvariablespredictingtomaternalreportof
boys’externalizingandinternalizingatage8
Maternalreportatage8(n¼206)
CBCLExternalizing CBCLInternalizing
Variables B Std.E b B Std.E b
Perinatalcomplications .223 .949 .029 .275 .918 .037
Unacceptingparenting 2.18 2.03 .286 )1.59 1.96 ).219
Familyadversityindex(FAI) 14.4 7.10 .287* 14.1 6.86 .293*
Perinatal·Parenting )1.17 .853 ).371 .189 .825 .063
Perinatal·FAI )1.55 2.97 ).093 )1.54 2.81 ).096
Parenting·FAI 9.65 5.55 ).459 3.27 5.37 .163
Perinatal·Parenting·FAI 5.48 2.48 .591* .178 2.39 .020
*p<.05.
Theinfluenceofperinatalcomplications 41
were set at or above the 75th percentile; low scores 4
were below the 75th percentile. Consistent with
3.5
previous research (Piquero & Tibbetts, 1999; Raine
et al., 1994, 1997), perinatal complications were 0 3
1
dichotomized, with zero complications comprising e
the ‘low group’ and 1 or more complications com- at ag 2.5
prising the ‘high group.’ As hypothesized, post hoc e 2
r
o
ANOVAs revealed that boys with high perinatal c
s
D 1.5
complications, high rejecting parenting, and high R
S
FAI demonstrated higher Externalizing than boys 1
with high scores on only 2 risk factors (61 vs.
0.5
51.33, respectively; F(1,58) ¼ 6.54, p < .05), 1 risk
factor (50.9; F(1,59) ¼ 6.07, p < .05), and no risk 0
No Risk Perinatal or FAI Risk Only Biosocial Risk
factors (50.29; F(1,113) ¼ 6.58, p < .05).
Table 4 shows the results for the hierarchical Figure1 Illustration of the significant interaction
regression analyses for child-reported delinquency detected between perinatal complications and family
at age 10, for which variables were entered into the adversity (FAI) predicting to boys’ self-reported delin-
regression equations as described above. Results quency(SRD) atage 10years
indicated a significant 2-way biosocial interaction
involving perinatal complications and family advers-
Discussion
ity. Post hoc contrasts conducted in the manner
described above revealed that boys with biosocial
The main goal of the present study was to test com-
risk reported more delinquent behaviors than boys
ponents of the biosocial model of early antisocial
with perinatal risk only (3.39 vs. 1.55, respectively;
behavior proposed by Raine (2002) in a sample of
F(1,161) ¼ 14.34, p < .001) and no risk (1.35;
low-income boys. A second aim was to test the spe-
F(1,66) ¼ 11.01, p < .001). There was also a trend
cificity of this model in relation to internalizing
for boys with biosocial risk to report greater de-
problems. Overall, there was broad support for the
linquency than boys with family risk only; however,
validity of the biosocial perspective in relation to
detectionofsuchaneffectwasdifficultbasedonthe
children’s antisocial behavior (ASB). Our findings
small number of cases in the family risk group
suggest that perinatal complications are related to
(n ¼ 4). Therefore, boys with family risk only were
ASB in boys but only in the context of compromised
combined with the group of boys demonstrating
family adversity or functioning. Addressing the spe-
perinatal risk only. When compared to the com-
cificity of the biosocial model, our findings did not
bined group of boys with only one risk factor, boys
support its generalizability to boys’ internalizing
with biosocial risk continued to report significantly
problems.Acrossreporters,thecumulativemeasure
higher rates of delinquency (1.52; F(1, 165) ¼ 15.1,
offamilyadversitywasaconsistentpredictorofboys’
p < .001) (see Figure 1).
later adjustment difficulties, supporting previous
findings and underscoring the salience of a negative
Internalizing psychosocial environment on childhood maladjust-
Tables 3and4displaytheresultsforthehierarchical ment (Sameroff et al., 1987; Shaw et al., 1994b,
regression analyses for maternal- and child-rated 1998a).
Internalizingatages8and10,respectively.Nosigni- The present study sought to replicate the findings
ficantinteractionswerefoundforeitheroutcome. and improve upon the methods in previous invest-
Table4 Summaryofregressionanalysesforperinatalcomplicationsandenvironmentalvariablespredictingtoyouthself-reportof
delinquencyandinternalizingatage10years
Youthself-reportatage10years(n¼196)
SRD–Delinquency CDI/MASC–Internalizing
Variables B Std.E b B Std.E b
Perinatalcomplications ).418 .214 ).240 ).100 .077 ).165
Unacceptingparenting .390 .456 .228 ).223 .164 ).376
Familyadversityindex(FAI) )2.32 1.60 ).206 .022 .576 .006
Perinatal·Parenting ).018 .192 ).025 .088 .069 .359
Perinatal·FAI 1.93 .655 .516** .218 .235 .169
Parenting·FAI )1.10 1.25 ).233 .718 .449 .440
Perinatal·Parenting·FAI .335 .557 .161 ).294 .200 ).408
*p<.05;**p<.01.
42 JoyE.BeckandDanielS.Shaw
igations (Arseneault et al., 2002; Raine et al., 1994, Anotherpossibleexplanationforthedifferencesin
1997). Both the Raine and Arseneault studies had the pattern of the interactions is that maternal
validated the importance of perinatal complications rejectionisamoresalientfactorfortheemergenceof
withinabiosocialframeworkinpredictingASBfrom ASBatearlierages(i.e.,age8versus10).Inprevious
early school age through adulthood, and this study research, a two-way biosocial interaction has been
corroborates these findings for ASB during middle found involving perinatal complications and meas-
childhood. In fact, the robustness of the interac- uresoffamilydisadvantageinrelationtoASBtaking
tional findings is noteworthy, such that boys place outside of the home (Arseneault et al., 2002;
experiencing high levels of perinatal, parenting, and Raine et al., 1994, 1997). This pattern of results is
family risk were described by their mothers as robust,forithasbeencorroboratedacrosstime(i.e.,
demonstrating ASB at least one standard deviation cohort effect), culture, and SES (Arseneault et al.,
above that of boys experiencing fewer risk factors 2002; Piquero & Tibbetts, 1999; Raine et al., 1994,
(d ¼ .99–1.08). Similarly, the effect sizes were high 1997;Werneret al.,1971);andourstudysupported
whenboysexperiencingbothperinatalriskandhigh theseresultswhenboysreportedontheirASBatage
familyadversitywerecomparedtoboysexperiencing 10. This is not to suggest that interactions involving
riskinonlyonedomainornoriskatall(d ¼ .82and parentingpersearenotaslikelytooccurinrelation
.85, respectively), according to youth report. Across to child ASB during the latter part of prepubescent
studies there appears to be convergence that boys period,butitmaybethecasethatothercomponents
withearlybiosocialriskdemonstratehigherlevelsof than rejection begin to play a more critical role as
ASB at various points in childhood, throughout children approach adolescence (e.g., monitoring).
adolescence, and into adulthood. The present study demonstrated support for a
The present findings also extend our understand- biosocial framework in predicting ASB but not
ing of possible mechanisms underlying interactions internalizing problems in childhood. This specificity
between perinatal complications and family advers- may be the results of differential consequences of
ityinrelationtoearlyASBfoundinpreviousstudies, perinatal complications for pathways leading to
by including more proximal family risk factors and externalizing versus internalizing problem behavior.
evaluating them more intensively. For example, For ASB, it is hypothesized that perinatal complica-
Raine and colleagues (1994, 1997) investigated the tions lead to neurological impairment in the frontal
interactiveeffectsofmaternalrejectionandperinatal lobes, an area of the brain believed to be dedicated
complications on adult males’ violent offending. Ra- primarily to executive functioning (Hawkins &
ther than using a direct measure of parenting after Trobsta, 2000; Raine, 1997). Deficits in executive
the child was born, these researchers employed functioning have been associated with problems in
three distal criteria (e.g., an attempt to abort the fe- impulsivity, verbal expression, and understanding
tus) to assess maternal rejection. Whereas in the societal norms, all validated correlates of ASB
present study, maternal parenting was evaluated (Raine, 2002). While children with internalizing
using observational strategies recorded in the home problemsmayexperiencesocialproblemsasaresult
and the laboratory. ofdeficitsinexecutivefunctioning(e.g.,withdrawing
Byinvestigatingthemoderatingeffectsofmultiple from their peers), social withdrawal may be less
environmentalfactors,ourstudyfoundasignificant impacted by harsh parenting and other types of
three-way interaction involving perinatal complica- family adversity. In contrast, for externalizing prob-
tions,rejectingparenting,andfamilyadversityinthe lems a child who suffers perinatal insult may be
prediction of ASB in boys at age 8. However, this especially sensitive to conditions of environmental
finding was only partially replicated at age 10. This risk. When reared in a noxious environment that
may be due, in part, to methodological considera- lacks sensitive and involved caregiving or that ex-
tions. First, different informants of child ASB were cludes the shaping and modeling of prosocial be-
used at the age-8 and age-10 assessments. Second, haviors, the child is not likely to develop these
childbehaviorislikelytohavebeenonlymoderately abilities. In all probability, coercive and aggressive
stable during this period. Third, although modestly behaviors will be learned instead, which sets the
correlated(r ¼ .15,p < .05),differentquestionnaires child up to fail within the school context, be it com-
were used to assess maternal- and youth-report of plying with teachers or socializing with peers (Pat-
ASB (CBCL and SRD, respectively), with only a terson, DeBaryshe, & Ramsey, 1989).
subset of these items included on both measures
(i.e., 60% of SRD items were covered on the CBCL
Limitations
Externalizing factor, whereas only 15% of CBCL
Externalizing items were represented on the SRD). A number of limitations must be considered when
Moreover, the CBCL Externalizing factor covers a interpreting the present results. First, many po-
broader domain of symptoms than the SRD, in- tentiatingfactorsofperinatalcomplications,suchas
cluding emotional control issues (e.g., jealous, lack of prenatal care, may be confounded with as-
stubborn/irritable, sudden mood changes, temper pects of family adversity. However, in this study
tantrums). perinatal complications and family adversity were
Theinfluenceofperinatalcomplications 43
unrelated(r ¼ .02),indicatingthattheseriskfactors impulsivity, high reactivity, excessively short atten-
reflectlargelyindependentprocesses.Second,CBCL tionspans),wouldbeadvocated.Inaddition,parents
dataandmanycomponentsoftheFAI(e.g.,parental insuchhigh-riskenvironmentsmayneedassistance
conflict, SES, parent criminality) were obtained in coping with stressors in and outside of the family
through maternal report, which raises the issue of thataffectthequalityofthecaregivingenvironment,
informant bias. However, many of the findings were as suggested by the results from analyses involving
corroborated by child-report, and perinatal compli- family adversity.
cations and maternal parenting were based on Insummary,thestudyprovidesdatathatindicate
objectivesourcesorobservation.Fourth,thepresent childrenexperiencingsevereperinatalcomplications
study was conducted on a high-risk sample of boys, are the most vulnerable to the effects of an adverse
which limits the generalizability of the findings to home environment. During prenatal care visits or
nonurban, less socioeconomically deprived popula- shortly after delivery, mothers who report an
tions.Moreover,theresultsofthisstudyoughttobe unstable home environment or frustration in caring
interpreted with caution in relation to predictors of for their older children would appear to be likely
girls’ externalizing behaviors and risk factors asso- candidates for intervention. Further prospective
ciated with other domains of functioning. For research of the interactive effects of perinatal insult
example, it is quite possible that results regarding and home environment on later child adjustment
internalizingproblemsmayvary,particularlyduring will help to clarify the precursors, reveal potential
earlyadolescencewhensexdifferencesindepression etiologies, and establish successful preventions of
emerge. early-onset ASB.
Finally, although this longitudinal study has
shown some support for the validity of the biosocial
perspectiveinregardstoboys’ASB,causalitycannot Acknowledgements
be inferred. Perhaps an uninvestigated third factor,
The authors are indebted to staff who assisted in
such as genetics, may be responsible for these re-
collecting and coding the data and to the research
sults (Raine, 2002). Child temperament and child
participants who made the study possible. This
psychopathology are often influenced by genetics.
research was supported by grants MH 46925 and
Negative parenting and a predisposition to perinatal
MH 50907 awarded to the second author from the
complications all contain genetic components, as
National Institute of Mental Health.
well. The present study cannot disentangle genetic
and environmental effects; twin and adoption
studies are needed to address this issue.
Correspondence to
Another ‘third variable’ explanation is ADHD.
Inattentiveness, hyperactivity, and impulsivity are JoyE.Beck,UniversityofPittsburgh,Departmentof
behaviors that tend to co-occur at high rates with Psychology, 210 S. Bouquet St., Rm 4425, Pitts-
ASB, and studies have demonstrated that boys with burgh, PA 15260, USA; Tel: [phone]; Fax:
comorbid ADHD and conduct disorder (CD) show a 412.624.8827; Email: [email protected]
more severe and chronic trajectory of antisocial be-
havior than CD-only and ADHD-only boys (Moffitt,
1990).Moreover,boyswithADHDshowlearningand References
social problems similar to those described in the
Achenbach,T.M.(1991).ManualfortheChildBehavior
sectionsabove.ToexplorethepossibilitythatADHD
Checklist 4/18 and 1991 Profile. Burlington, VT:
symptoms accounted for much of the variance
University of Vermont, Department ofPsychiatry.
attributed to perinatal complications, we entered
Aguilar, B., Sroufe, A., Egeland, B., & Carlson, E.
ADHD as a covariate into the multiple regression
(2000).Distinguishingtheearly-onset/persistentand
equationsanddidnotfindADHDtoaffectanyofthe
adolescent-onset antisocial behavior types: From
significant interactions predicting to ASB. birthto16years.DevelopmentandPsychopathology,
12, 109–132.
Arseneault, L., Tremblay, R.E., Boulerice, B., & Sau-
Clinical implications and conclusions
cier, J. (2002). Obstetrical complications and violent
A major implication of this study is the import of delinquency: Testing two developmental pathways.
early environmental risk factors when considering Child Development,73, 496–508.
Bost,K.K., Vaughn, B.E.,Washington,W.N., Cielinski,
later adjustment of high-risk children. Most aspects
K.L., & Bradbard, M.R. (1998). Social competence,
of risk in this study were measured at or before the
social support, and attachment: Construct domains,
age of 2. Thus, one implication is that preventative
measurement, and paths of influence for preschool
interventions could target toddlers from disadvant-
children. ChildDevelopment, 69,192–218.
aged or chaotic families to maximize benefits on
Brennan, P., & Mednick, S.A. (1997). Medical histories
child outcome. Efforts to teach parenting skills, of antisocial individuals. In D.M. Stoff, J. Breiling, &
especially for toddlers showing extreme signs of J. Maser(Eds.), Handbookofantisocial behavior(pp.
compromised executive functioning (e.g., constant 269–279). NewYork:John Wiley & Sons.
44 JoyE.BeckandDanielS.Shaw
Broman, S.H., Nichols, P.L., & Kennedy, W.A. (1975). Academy of Child and Adolescent Psychiatry, 39,
PreschoolIQ:Prenatalandearlydevelopmentalcorre- 1229–1237.
lates. Hillsdale,NJ: Erlbaum. March, J., Parker, J., Sullivan, K., Stallings, P., &
Caldwell,B.M.,&Bradley,R.H.(1984).HomeObserva- Conners, C.K. (1997). The Multidimensional Anxiety
tionforMeasurementoftheEnvironment.LittleRock, Scale for Children (MASC): Factor structure, reliabil-
AK: Universityof Arkansas. ity, and validity. Journal of the American Academy of
Campbell, S.B., Pierce, E.W., Moore, G., & Marakovitz, Child andAdolescentPsychiatry, 36, 554–565.
S.(1996).Boys’externalizingproblemsatelementary Martin, J. (1981). A longitudinal study of the conse-
school age: Pathways from early behavior problems, quences of early mother–infant interaction: A micro-
maternalcontrol,andfamilystress.Developmentand analytic approach. Monographs of the Society for
Psychopathology, 8,701–719. Research inChild Development,46 (SerialNo. 190).
Campbell,S.B.,Shaw,D.S.,&Gilliom,M.(2000).Early Mednick,B.R.,Hocevar,D.,Baker,R.L.,&Teasdale,T.
externalizing behavior problems: Toddlers and pre- (1983). Effects of social, familial, and maternal state
schoolers at risk for later maladjustment. Develop- variables on neonatal and infant health. Develop-
ment andPsychopathology,12, 467–488. mentalPsychology, 19, 752–765.
Cicchetti, D., & Rogosch, F.A. (1996). Equifinality Moffitt, T.E. (1990). Juvenile delinquency and atten-
and multifinality in developmental psychopathology. tion-deficitdisorder:Developmental trajectoriesfrom
Development andPsychopathology, 8, 597–600. age 3 to15. Child Development, 61, 893–910.
Cohen, P., Velez, C., Brook, J., & Smith, J. (1989). Moffitt, T.E. (1993a). Adolescent-limited and life-
Mechanisms of the relation between perinatal prob- course-persistentantisocialbehavior:Adevelopmen-
lems,earlychildhoodillness,andpsychopathologyin taltaxonomy. PsychologicalReview, 100,674–701.
late childhood and adolescence. Child Development, Moffitt, T.E. (1993b). The neuropsychology of conduct
60, 701–709. disorder. Development and Psychopathology, 5, 135–
Crnic,K.A.,&Greenberg,M.T.(1990).Minorparenting 151.
stresseswithyoungchildren.ChildDevelopment,61, Patterson, G.R., DeBaryshe, B., & Ramsey, E. (1989).
1628–1637. A developmental perspective on antisocial behavior.
Deater-Deckard, K., Dodge, K.A., Bates, J.E., & Pettit, American Psychologist, 44,329–335.
G.S. (1998). Multiple risk factors in the development Piquero, A., & Tibbetts, S. (1999). The impact of pre/
of externalizing behavior problems: Group and in- perinatal disturbances and disadvantaged familial
dividual differences. Development and Psychopatho- environmentinpredictingcriminaloffending.Studies
logy, 10, 469–493. onCrime andCrimePrevention, 8, 52–70.
Dishion, T.J. (1990). The family ecology of boys’ peer Pittsburgh Youth Study. (1991). Neighborhood danger-
relationsinmiddlechildhood.ChildDevelopment,61, ousness scale. Unpublished measure, University of
874–892. Pittsburgh.
Dodge, K.A., Pettit, G.S., & Bates, J.E. (1994). Social- Raine, A., Brennan, P., Mednick, B., & Mednick, S.
ization mediators of the relation between socioeco- (1996). High rates of violence, crime, academic
nomic status and child conduct problems. Child problems, and behavioral problems in males with
Development, 65, 649–665. both early neuromotor deficits and unstable family
Drillien, C.M. (1964). The growth and development of environments. Archives of General Psychiatry, 53,
the prematurely borninfant. Edinburgh:Livingstone. 544–549.
Elliott, D.S., Huizinga, D., & Ageton, S.S. (1985). Raine, A. (1997).Antisocial behavior andpsychophysi-
Explaining delinquency and drug use. Thousand ology: A biosocial perspective and a prefrontal dys-
Oaks, CA: Sage Publications. function hypothesis. In D.M. Stoff, J. Breiling, & J.
Hawkins, K.A., & Trobsta, K.K. (2000). Frontal lobe Maser (Eds.). Handbook of antisocial behavior
dysfunction and aggression: Conceptual issues and (pp.289–304). NewYork: JohnWiley & Sons.
researchfindings.AggressionandViolentBehavior,5, Raine, A. (2002). Biosocial studies of antisocial and
147–157. violent behavior in children and adults: A review.
Kandel, E., & Mednick, S. (1991). Perinatal complica- Journal of AbnormalChild Psychology, 30,311–326.
tions predictviolent offending.Criminology, 29,519– Raine, A., Brennan, P., & Mednick, S. (1994). Birth
529. complications combined with early maternal rejec-
Kazdin, A.E. (1987). Conductdisorder inchildhood and tions at age 1 year predispose to violent crime at age
adolescence.Newbury Park,CA: Sage Publications. 18 years. Archives of General Psychiatry, 51, 984–
Kazdin,A.E.,French,N.H.,Unis,A.S.,Esveldt-Dawson, 988.
K.,&Sherick,R.B.(1983).Hopelessness,depression, Raine, A., Brennan, P., & Mednick, S. (1997). Interac-
and suicidal intent among psychiatrically disturbed tion between birth complications and early maternal
inpatient children.Journal ofConsultingandClinical rejection in predisposing individuals to adult viol-
Psychology, 51, 504–510. ence: Specificity to serious, early-onset violence.
Kovacs,M.(1992).TheChildren’sDepressionInventory American Journal ofPsychiatry, 154,1265–1271.
Manual. North Tonawanda, NY: Multi-Health Sys- Rantakallio,P.,Koiranen,M.,&Moettoenen,J.(1992).
tems. Association of perinatal events, epilepsy, and central
Laucht,M.,Esser,G.,Baving,L.,Gerhold,M.,Hoesch, nervous system trauma with juvenile delinquency.
I., Ihle, W., Steigleider, P., Stock, B., Stoehr, R., Archives of DiseaseinChildhood,67, 1459–1461.
Weindrich, D., & Schmidt, M. (2000). Behavioral Richardson,G.A.,Hamel,S.C.,Goldschmidt,L.,&Day,
sequelaeofperinatalinsultsandearlyfamilyadvers- N.L. (1999). Growth of infants prenatally exposed to
ity at 8years of age. Journal of the American cocaine/crack: Comparison of a prenatal care and a