Introduction

DSM-IV TR describes the two Disruptive Behavior Disorders (DBD), Oppositional Defiant Disorder (ODD) and Conduct Disorder (CD), as distinct, casting them in a hierarchical fashion, with CD distinguished from ODD by a greater severity of antisocial behavior (American Psychiatric Association 2000). Despite much overlap, modest evidence in general population and case-control samples supports this distinction. Most research addressing this issue has examined child-related correlates and found group differences in diagnostic comorbidities (Burke et al. 2005; Cohen and Flory 1998; Maughan et al. 2004), as well as similarities in demographics (e.g., age, gender; Rowe et al. 2002).

Given the heterogeneity in the behaviors that comprise ODD and CD, investigators have sought to identify other distinguishing features to improve models that disentangle multiple pathways to CD. Early conduct problems have been identified as the most common cause of referrals for treatment among children and adolescents (see Nock et al. 2006), although the degree to which service seeking is due to ODD versus CD specifically is not clear. ODD is regarded as a key risk factor for CD, yet only a minority of children with ODD will develop CD (Loeber et al. 2000a). Thus, in terms of prognosis and selection of treatment strategies, clinicians have a limited evidence base upon which to judge the likelihood that a given child will progress from ODD to CD. Some evidence suggests that certain risk factors in contextual domains (e.g., parenting problems, parent-child conflict, deviant peers) may convey greater risks for CD or delinquency than for ODD, although the current evidence is not entirely clear (Burke et al. 2002; Loeber et al. 2000b). One general population study did find that both child characteristics (e.g., age, gender) and contextual problems showed stronger associations to antisocial, than oppositional, behavior (e.g., parental criminality, parental substance problems, adverse parenting; Rowe et al. 2002).

Another goal for clinical practice is to ultimately improve the efficient use of clinical resources in treating DBDs. Although multimodal interventions targeting several domains are typically more effective than individually focused interventions (Burke et al. 2002), the available literature has not adequately identified which specific contextual factors exert the greatest influence on the worsening progression of conduct problems. The identification of these contextual predictors is crucial not only to enable the individual clinician to weigh the relative importance of specific risk factors across multiple domains for a given child, but also to promote the broader development and evaluation of more efficiently targeted multimodal intervention strategies. For instance, very few options are available to incorporate peer or neighborhood factors in treatment, even though risk factors in these domains may complicate the effectiveness of treatment in a particular case. Vitaro et al. (1999) found that reductions in both disruptive behavior and associations with less deviant peers mediated the reduction in rates of subsequent CD following intervention. Likewise, antisocial behavior has been linked with school behavior and performance problems, but this relationship may be influenced by other background variables (Hinshaw 1992). Finally, disadvantaged neighborhoods have been found to moderate the impact of parenting and sibling effects on children’s conduct problems (Brody et al. 2003). Thus, in order to identify key elements to be targeted during interventions for ODD and CD, and to guide the development of appropriate interventions, it is imperative to improve existing descriptions of the contextual influences on antisocial behavior.

Only a few studies have compared the contextual correlates of CD and ODD in clinically referred children. Early reports found that parental psychopathology and substance abuse were more related to CD (Frick et al. 1991) than ODD (Frick et al. 1992). Deviant or inappropriate parenting practices have been more strongly associated with CD than ODD in some studies (Burke et al. 2003; Lindahl 1998), but not others (Frick et al. 1992; Green et al. 1992). Using one of the largest clinically referred samples, Greene and colleagues (2002) found that ODD+CD (vs. ODD alone) was associated with child-related characteristics (e.g., males, lower adaptive functioning, depression) and one parent or family and school variable (lower socioeconomic status, special class placement), but did not differ in age, ADHD, and family functioning.

A major limitation in understanding the contextual correlates of ODD with and without CD is the relative absence of direct comparisons of these disorders on measures across key contextual factors (e.g., parent, family, peer, school, community) that are related to DBDs (Loeber et al. 1998) and the developmental progression towards more serious antisocial behavior (Babinski et al. 1999; Broidy et al. 2003). Many of these factors have been postulated as important in lifespan developmental models of DBDs (see Loeber et al. 2000a). Relatedly, most of the early studies did not report whether children diagnosed with CD also met research criteria for comorbid ODD, nor did they include multiple measures from several contextual domains, which could have significantly limited their results. Further, early studies generally included only boys (e.g., Frick et al. 1992; Frick et al. 1991; Green et al. 1992). Given increased attention to antisocial behavior in girls (Pajer 1998) and recent findings suggesting similar social and family risks for DBDs in both sexes (Messer et al. 2006), whether younger girls experience problems in contextual domains comparable to those of younger boys is an important issue to examine. Finally, there is a need to determine if ODD children with or without CD differ on contextual variables from healthy controls, given the relative absence of studies that include all three groups. Comparisons of all three groups would clarify the extent to which contextual problems serve as critical foci for prevention or intervention among DBD children, and, if contextual differences are found between the two DBD groups, whether interventions should be tailored to reflect these differences. The use of healthy control groups selected for the absence of psychopathology has been found helpful in other studies comparing disordered and non-disordered samples (Birmaher et al. 2004; Williamson et al. 2004).

The present study extends prior research by comparing young, clinically referred boys and girls diagnosed with ODD + CD or ODD alone to one another and to a third group of matched, healthy controls (HC) on variables from key contextual domains. In contrast to prior studies, the study incorporates DSM-IV criteria in the context of structured clinical interviews conducted with children and parents using a psychometrically sound measure. Because of the need to examine psychological, biological and contextual factors that may contribute over time to the development of the hierarchically-related disorders ODD and CD, it was important to select a comparison group that was unlikely to later meet criteria for a DBD. To avoid potential group confounds, we chose a control group that did not evidence current or past psychopathology in an effort to reduce the likelihood that these children would develop any behavioral problems, in particular. We explore the following questions: (1) How do the two DBD groups (ODD + CD, ODD alone) differ from HCs in various contextual characteristics (e.g., family conflict, peer delinquency)?; (2) how do children who meet criteria for ODD + CD differ from children who meet criteria for ODD alone on the same contextual characteristics?; and (3) do the findings in the full sample parallel those found for girls only. The study is novel in its incorporation of these diagnostic subgroups by including diverse constructs in several contextual domains related to antisocial behavior, multiple informants, clinically referred boys and girls, a matched healthy control group, and multivariate models.

Method

All Participants

Participants were 176 clinically referred children diagnosed with either ODD only (n = 136) or ODD + CD (n = 40), and 69 matched healthy controls (HC). All met the following inclusion criteria: (1) boys or girls aged 6–11 years, (2) residence with at least one parent/guardian, (3) intellectual level no more than two SDs below age norms (Kaufman and Kaufman 1990), and (4) parent consent and child assent for participation as approved by the University’s Institutional Review Board. Cases were excluded for any of the following criteria: (1) current participation in treatment; (2) psychosis, bipolar disorder, major depression with vegetative signs, substance abuse disorder, or an eating disorder; or (3) suicidality with a plan or any homicidality.

Overall, the child sample had an average age of 9.0 years (SD = 1.7), was mostly boys (81%), and had a comparable proportion of Caucasian (53%) and African-American or biracial (47%) children. In terms of the adult informants, all but one was the primary caregiver. A total of 211 (86.1%) informants were biological mothers, 22 (9.0%) were biological fathers, and 12 (4.9%) consisted of other relatives and adoptive parents. A total of 106 adults (60%) were married, and 164 (66.9%) adults had at least some college experience. Overall, 127 (51.8%) of the children lived in single-adult households.

Samples/Recruitment Sites

Clinically Referred Sample with DBD

The DBD sample (n = 176) consists of 138 children who were clinically referred to this study for randomization into one of two outpatient treatment protocols and another 38 clinically referred children who were already scheduled to receive existing outpatient services through programs affiliated with the University of Pittsburgh Medical Center. All of these cases were referred primarily by schools (35%), other UPMC clinics (31%), advertisements (18%), other parents (8%), outside agencies (4%), and unknown sources (2%). The 138, protocol-randomized cases were recruited through advertisements or brochures sent to schools and local mental health centers. The 38 existing services cases were recruited at or immediately after their intake appointment at one of two outpatient child/adolescent clinics, in order to provide a comparison to “treatment as usual”. Relative to the existing services cases, the protocol cases were more likely to be boys, χ2 (1, N = 185) = 5.62, p < .02, with higher composite IQ scores, F (1, 184) = 5.05, p < .05, and higher family socioeconomic status [SES], F (1, 184) = 21.51, p < .0001.

Healthy Control (Nonpatient) Sample

Recruited to provide “normal comparisons,” the healthy control (HC) sample consisted of children who did not meet diagnostic criteria for a current or past DSM-IV disorder, though they may have met several subthreshold criteria for various symptoms of a given diagnosis. The group did show variability in T-scores on the CBCL externalizing scale (range: 30–62). HCs were recruited through advertisements in newspapers and flyers posted in children’s centers. To obtain a demographically comparable sample, HCs were matched to a randomly selected subset of 40 children from the protocol-randomized sample and 20 children in the existing services sample based on four background criteria: (1) age (±6 months), (2) sex, (3) race/ethnicity, and (4) family SES (within 10 points on the Hollingshead; Hollingshead and Redlich 1958). We found complete matches on all four criteria for 87% of these children. However, given difficulties finding full matches within the study time frame, we included nine additional (13%) HCs with ages matched within nine months and three (4%) cases with SES matched within 12 points.

Diagnostic Subgroups

The three diagnostic subgroups were similar in age, sex, and racial background. Relative to the HC group, the ODD + CD and ODD only groups had lower composite IQs (107.3 vs. 95.6, 100.1), F (1, 184) = 5.05, p < .05, and family SES scores (38.5 vs. 30.8, 34.2), F (1, 184) = 5.05, p < .05, but the groups still fell into the same SES category. The proportion of informants who were self-identified as the child's mother were comparable in the ODD + CD (87.2%), ODD only (85.4%), and healthy control groups (87.0%), χ2 (2, N = 245) = 3.7, p < 0.45. The ODD + CD (67%) and the ODD only groups (53%) did have a higher proportion of single-adult households than the healthy control group (42%), χ2 (2, N = 245) = 6.12, p < .05). Regarding psychiatric comorbidity, the ODD + CD and ODD only groups had high but comparable proportions of cases with ADHD (80.0% vs. 73.5%), χ2 (1, N = 176) = 0.51, p < 0.53, or a non-DBD disorder, 51.3% vs. 36.5%; χ2 (1, N = 176) = 2.83, p < 0.14.

Procedures

Besides an evaluation of the child’s psychiatric disorders, the intake assessment incorporated multiple informants (e.g., child, parent, teacher), methods (e.g., self-reports, interviews), and measures evaluating child functioning and several domains relating to the context in which DBDs emerge. Diagnostic interviews were conducted in a project office by a master’s-level clinician. Other research assessments were conducted with the child and primary caregiver (85% mothers) by separate, trained research assistants. Child and parent informants in the DBD groups were paid $10 each for completing measures at intake. The two informants in the HC group were paid $40 each. The HC group was given a larger amount of compensation given that they were volunteers for the study who would otherwise garner no additional benefits to study participation.

Measures of Child Functioning

Diagnosis/Comorbidity

The Schedule for Affective Disorders and Schizophrenia for School-Aged Children for DSM-IV—Present and Lifetime (K-SADS; Kaufman et al. 1996; v. 10/1/96) diagnostic interview was administered to the child and parent to evaluate presence of child psychiatric disorders. The procedure involved separate parent and child interviews, followed by an interview with both informants together to reconcile differences for ratings of individual symptoms. Inter-rater reliability was examined based on ratings obtained from a trained interviewer who viewed videotapes of initial interviews. The reliabilities (n = 57) were moderate-to-high for disorders occurring more than once in this sample, namely, ADHD (k = .77), ODD (k = .70), CD (k = .76), Major Depression (k = 1.00), and Anxiety Disorder (k = .83). Since diagnostic groups were selected based on ODD and CD, we compared groups on presence of ADHD and other non-DBD diagnoses.

Attention Problems

The primary caretaker completed the Child Behavior Checklist (CBCL) to assess behavioral and emotional problems, and social competencies (Achenbach 1991). In this study, we used the attention problems score as a covariate in our group comparison analyses, given its strong comorbidity with DBD status.

Contextual Variables Related to DBD

Parental Dysfunction

We included an aggregate of the five subscales from the Parental Self-Efficacy Scale (PSES; Evans et al. 1997) to evaluate overall level of self-confidence in key parental roles (i.e., emotional support, behavior management; advocacy, provider issues, school issues, α = .75). Caregiver ratings (1–4 pt. scale) reflect overall comfort in ability to perform these tasks (e.g., discipline child when necessary). We also reported the hostility scale score from the Brief Symptom Inventory (BSI; α = .75) reflecting various forms of psychological distress during the past week (Derogatis et al. 1976), given the relationship of this construct to antisocial behavior (Lahey et al. 1989). Finally, parents also completed the 21-item Beck Depression Inventory (BDI; Beck et al. 1988). The BDI has excellent psychometric properties (α = .88).

Parenting Practices

Children completed the Parent Perception Inventory (PPI) to evaluate the primary caretaker’s overall practices (Hazzard et al. 1983). The PPI yields a summary score reflecting level of positive behavior minus negative behavior (α = .80). The Alabama Parenting Questionnaire (APQ; Shelton et al. 1996) was administered to parents to evaluate six dimensions of parenting practices related to antisocial behavior, yielding summary scores reflecting positive (e.g., involvement) and negative (poor monitoring) practices (α’s = .76, .80). We also included the Parent–Child Conflict Tactics Scale (PCCTS; Straus et al. 1998) focusing on the 5-item psychological aggression factor (α = .50), which was selected to complement the APQ. Parental reports of the overall frequency with which they engaged in each behavior are assigned a category score and then summed across items (e.g., shouted, yelled, or screamed).

Family Functioning/Stress

Children were administered the 11-item family support scale of the Survey of Children’s Social Support (1–5 pt. scale ratings) to provide an indicator of self-reported level of support (α = .70) in the family domain (SOCCS; Dubow and Ullman 1989). Parents completed the Family Adaptability and Cohesion Evaluation Scales (FACES-II) to provide an index of cohesion and adaptability (Olson et al. 1982). The adaptation (23 items; α = .76) and cohesion (27 items; α = .87) scales are rated on 1–5-pt. Likert scales. The Life Events Checklist (LECL) was administered to parents to evaluate the child’s exposure to 46 stressful life events in the past year (Brand and Johnson 1982), yielding an overall life events impact score reflecting the sum of identified positive and negative life events.

Peer Relations

To examine peer relations, children completed the 15-item peer support scale of the aforementioned SOCSS (Dubow and Ullman 1989). The SOCSS has strong psychometrics with elementary-school-aged children (α = .80). The Peer Delinquency Scale (PDS; Elliot et al. 1985) was administered to evaluate the child’s exposure to delinquent peers. We report the percentage of endorsed peer involvement in 10 delinquent acts (α = .92).

School Functioning

To examine the child’s functioning at school, we report the externalizing and academic performance scales from the Teacher Report Form (TRF; Achenbach 1991) to provide an index of behavior problems and productivity at school. In paralleling the parent-reported CBCL, the teacher-completed TRF is based on normative T-scores (M = 50; SD = 10). Because of its lower sample size relative to other measures, the TRF was not included in any overall multivariate analyses.

Community Characteristics

Parents completed an abbreviated version of the Bad Neighborhood questionnaire to screen for the child’s exposure to deviance in their immediate community (BN; Loeber et al. 1998). We combined the original 17 items to form five general items (e.g., crime, drug use/dealing, gangs/transients/street people) to provide basic information regarding adverse or dangerous neighborhood influences related to CD (α = .92). Children completed ratings on the 5-item teacher support scale (α = .66) of the SOCSS.

Data Analysis Approach

To reduce the number of variables examined across contextual domains, some individual variables from the same measures were aggregated into constructs. We ran separate confirmatory factor analyses with the prevailing factor scores from each measure to determine which variables fit the data well and should be retained for analysis (see Eid et al. 2006). We retained all variables that met acceptable selection criteria (i.e., Comparative Fit Index ≥0.95 and Root Mean Square Error of Approximation ≤.05). Our analytic strategy was modeled after that reported by Greene et al. (2002) in their comparison of children with ODD + CD, ODD only, or neither disorder (controls). Accordingly, we conducted an overall multivariate analysis of covariance (MANCOVA) with all three groups to examine all of the variables within a given domain. If the overall test was statistically significant, we then conducted an analysis of covariance (ANCOVA) with all three groups using each variable in the same domain, which, if significant, was followed by pairwise comparisons. Both sets of analyses controlled for overall initial group differences on three key background variables (child IQ, CBCL attention problems scale, family SES), in order to be consistent with other studies comparing DBD to non-DBD samples that have adjusted for background characteristics that include SES. We also conducted a separate set of exploratory analyses using girls only. Due to low number of girls (n = 46), we compared the two DBD groups combined to the HC group. We then tested a final multivariate model that entered sets of variables in two blocks: (1) the two background variables alone (block 1), and (2) the two background variables plus all significant contextual variables (block 2). We sought to include the child attention problems scale in block 3, but the model was overwhelmed upon its inclusion. These variables were entered into a logistic regression using forward and then backwards entry designed to differentiate each group from one another (ODD + CD vs. HC; ODD only vs. HC; ODD + CD vs. ODD only). The minimum criterion for entry for each variable was p < .05. Odds ratios (ORs) and 95% confidence intervals (CIs) were computed for significant predictors. All analyses were conducted using SPSS (v. 14.0). Baseline levels of missing data across measures were very low for the child and parent informants (M = 1.4%).

Results

Comparisons Among the ODD + CD, ODD only, and HC Groups on Contextual Measures

Table 1 displays descriptive statistics and the results of MANCOVAs examining overall group differences, and follow-up ANCOVAs, for variables in each of the five contextual domains. For each set of pairwise comparisons, we reported a significance level and an effect size to depict the nature of any group differences. Omnibus tests comparing the three groups on the measures contained in each domain were significant, except for community characteristics. No further follow-up tests were conducted for this domain as the three groups showed comparable levels of exposure to community deviance and perceived teacher support. Thus, for each domain, we reported any univariate differences between the two DBD groups and the HC group, and any differences between the ODD + CD and ODD only groups.

Table 1 Comparisons of the ODD + CD, ODD only, and healthy control groups on contextual variables in different domains
Full size table

Parental Dysfunction

The ODD + CD and ODD only groups each differed significantly from the HC group in level of parental self-efficacy, but only the ODD + CD group differed from the HC group in self-reported hostility. The two DBD groups and HC group did not differ in mean severity of parental depression. Parents of ODD + CD children reported greater hostility than parents of ODD only children.

Parenting Practices

The ODD + CD and ODD only groups reported less positive parenting and greater use of psychological aggression than the HC group. The two DBD and HC groups were not found to differ in parent-reported negative parenting and child-reported overall negative-to-positive parental management methods. Parents of ODD + CD and ODD only children did not differ significantly on any of these measures.

Family Functioning

The DBD groups reported more stressful life events than the HC group, but only the ODD only group had a significantly higher level than the HC group; the three groups did not differ in reports of family support, cohesion, or adaptability. The two DBD groups did not differ on any of these measures.

Peer Relations

The two DBD groups had significantly greater affiliations with delinquent peers than the HC group, but the groups did not differ in perceived peer support. ODD + CD children reported higher levels of affiliations with delinquent peers than ODD only children, but the difference did not reach statistical significance. These two groups also reported comparable levels of perceived peer support.

School Functioning

The ODD + CD and ODD only groups had significantly greater externalizing behavior problems at school than the HC group, but they did not differ in teacher ratings of their academic performance. The two DBD groups were reported as being comparable in their levels of externalizing behavior problems and academic performance.

Multivariate Models Comparing Healthy Controls to the ODD + CD or ODD Only Group

In the full sample, we used logistic regression to examine final multivariate models for distinguishing each of the DBD groups (ODD + CD, ODD only) from the HC group, using variables selected on the basis of the previous univariate tests. Due to lower sample sizes, we did not include school variables. The logistic regression analysis that sought to identify the correlates of ODD + CD (vs. HC) using the full sample included one peer, two parental dysfunction, and two parenting practices variables, in addition to the two background characteristics (SES, IQ). In Table 2, three variables remained in the model with narrow 95% confidence intervals (CI), suggesting that the ORs were robust and accurate, χ2 (5, N = 102) = 103.71, p < .0001. As expected, ODD + CD status was associated with greater parental hostility and exposure to delinquent peers, and less parental self-efficacy, than HC status. When these variables were entered in the second step, SES and IQ were non-significant contributors to subgroup classification.

Table 2 Multivariate models examining the contextual correlates of ODD + CD and ODD only
Full size table

A separate logistic regression examining the correlates of ODD only (vs. HC) included the same two background characteristics and all five contextual variables as used in the preceding analysis, except one (parental hostility). As shown in Table 2, ODD only status was associated with greater parental use of psychological aggression and child exposure to peer delinquency, lower parental self-efficacy, and fewer overall positive-to-negative family life events, χ2 (6, N = 203) = 195.42, p < .0001. The 95% CIs for all four contextual variables were narrow in range. The two child demographics were non-significant contributors to the model when added in the second step.

Comparisons of the Two DBD Groups Combined and HC Group in Girls Only

Using the subset of girls (n = 46), we conducted exploratory MANCOVAs and follow-up ANCOVAs comparing the DBD (ODD + CD and ODD only combined) and the HC groups to determine whether the pattern of findings with girls would parallel the pattern found in the full sample. The small sample of ODD + CD girls (n = 6) precluded comparison of the two DBD subgroups. As shown in Table 3, the girls with DBD generally showed greater problems in parental dysfunction, parenting practices, and peer relations. Specifically, DBD girls had parents who reported heightened personal hostility and greater use of psychological aggression, and who reported less parental self-efficacy. DBD girls also reported less positive parenting and greater exposure to delinquent peers. The means on the two school functioning variables for the DBD and HC girls were in the right direction, but were not significantly different from one another. As with the full sample, there were also no differences in the family functioning or community status of the DBD and HC girls.

Table 3 Comparisons of girls in the DBD and healthy control groups on contextual variables in different domains
Full size table

Discussion

This study compared clinically referred children diagnosed with DBD to a matched sample of healthy controls and compared those who met criteria for ODD + CD vs. ODD only on measures evaluating key constructs across contextual domains. The study is unique in its inclusion of clinically referred boys and girls, diverse constructs and informants, and healthy controls. The two DBD groups differed from HCs on five of six domains, but those with ODD + CD or ODD only were found to be comparable on all but one measure across the same domains. Findings were similar when the analysis was restricted to a comparison of DBD vs. HC girls. Further, the overall findings were not accounted for by key child background variables.

In the full sample, the ODD + CD and ODD only groups differed from the HC group on some, but not all, of the selected contextual measures. Specifically, the DBD groups were characterized by certain forms of parental dysfunction (hostility) and parenting practices (fewer positive, but not more negative practices), including a lowered sense of parental self-efficacy and greater use of psychological aggression. Further, DBD children lived in families with exposure to more negative family life events, had greater school behavior problems, and were more exposed to delinquent peers. At the same time, the DBD groups and HC group were similar on several other contextual variables, such as parental problems (depression) and parenting practices (use of negative parenting practices), family support and adaptability, peer and teacher support, and neighborhood adversity. These findings provide support for the role of certain contextual variables as potential risk factors for the development of a DBD in general (e.g., Lahey et al. 1999), rather than being specific to having CD (e.g., Frick et al. 1992). Our findings indicating greater differences between the DBD and HC groups on parental dysfunction and parenting behaviors, and few differences on family and community factors, extend previous work suggesting that parenting behaviors may influence the effects of environmental factors on children’s behavioral problems (Loeber et al. 2000a). The lack of association between DBD status and community context may be difficult to interpret given the limited nature of our measures in this domain. It is also possible that these early-onset disorders in childhood may not yet have caused disruptions in the community domain, or are less associated with these variables, which could imply that disruptions are an outcome rather than a cause of DBDs.

In separate multivariate models, HCs differed from both the ODD + CD and the ODD only groups on two variables, less parental self-efficacy and greater exposure to delinquent peers. Parental hostility further differentiated the CD + ODD and HC groups, while parental use of psychological aggression and more negative life events further differentiated the ODD only and HC groups. In addition, the fact that parental hostility was related to CD + ODD only, whereas parental use of psychological aggression was related to ODD only, may highlight unique contributors to each of these subgroups. Certainly, the processes by which DBDs and parental practices develop are very likely to be interactive and reciprocally influential (Lahey et al. 1999).

Our exploratory analyses comparing DBD and HC girls paralleled the analyses with the full sample on 12 of 17 measures (e.g., greater parental hostility and psychological aggression, less parental self-efficacy and positive parenting, more peer delinquency). In contrast to the analyses using the full sample, girls with DBD reported more negative parenting practices than HC girls, and their parents did not differ in reports of positive parenting or negative life events. Also, the DBD and HC girls’ levels of school behavior problems and academic performance differed by a full standard deviation, though the differences were not significant. Collectively, these similarities may suggest the relevance of multimodal intervention strategies to both sexes, with perhaps some greater attention to parenting practices and peer interventions for girls (see Messer et al. 2006; see Webster-Stratton 1996).

The contextual characteristics of ODD + CD and ODD only children were found to be remarkably similar and they differed on only one variable. Specifically, ODD + CD children had parents who reported greater hostility and they tended to report greater exposure to delinquent peers, than ODD only children. Otherwise, there were similarities in parental practices, family dysfunction, and community characteristics. In addition, the two DBD subgroups did not differ in comorbid ADHD. Heightened antisocial behavior has been related to parental hostility (Rey and Plapp 1990) and exposure to peer delinquency (Burke et al. 2005), and there is some evidence to suggest that hostility is related more to CD than ODD (Lahey et al. 1989). One implication of these findings is that the positive and negative contextual processes assessed herein do not appear to account for a child’s progression to more serious antisocial behavior. It is also possible that these contextual factors may interact with other child-related characteristics or they exert greater influence in an older child sample, which we plan to examine.

Overall, the current findings indicate that poor parental functioning, parent management problems, exposure to deviant peers, and school behavior problems are among the more robust contextual factors that may influence the development of ODD and early-onset CD, consistent with other studies (see Lahey et al., 1999; Loeber et al. 2000a; Patterson et al. 2000). At the same time, only the level of parental hostility clearly discriminated ODD children with or without comorbid CD (Burke et al. 2002), suggesting that these contextual variables are unlikely to account for the developmental progression from ODD to CD as described in the DSM-IV (Loeber et al. 2000a, b). Our findings were not due to differences in age, sex, or intellectual functioning, in accord with prior work showing similar symptom utility patterns for ODD and CD children across age and sex groups (see Loeber et al. 1998). Additionally, the two groups did not differ in the proportion of children with ADHD or internalizing problems, suggesting that such features may be more reflective of DBD as a whole. Since few studies have compared ODD with or without CD on such measures, further evidence is needed to document the generalizability of our findings and how contextual correlates influence change in ODD or CD symptoms or genetic risk factors for antisocial behavior (see Burke et al. 2002).

Among the study’s limitations, the small sample of ODD + CD children may have limited power to detect group differences, and the use of clinical referrals vs. non-referred children (see Beauchaine et al. 2001; Frick et al. 1991; Lindahl 1998) could have minimized the likelihood of finding group differences between ODD + CD vs. ODD only. As was evident for both DBD groups, parents who seek services for their children may be more similar in their levels of various contextual problems, such as parental distress or family dysfunction (Lahey et al. 2002), which may reflect an increased severity threshold necessary to motivate parents to seek treatment. It is also possible that the CD observed in referred and non-referred populations may be quite different from one another. In non-referred populations, CD occurs without ODD much more frequently (e.g., Maughan et al. 2004; Rowe et al. 2002; Simonoff et al. 1997) than in clinic populations where CD usually occurs with ODD (e.g., Biederman et al. 1996; Greene et al. 2002; Lahey et al. 1995). Of course, it is important for our purposes to understand how those children who progress from ODD to CD differ from those who do not show this progression, which may help to focus intervention efforts when a child presents with ODD. This suggests the need to conduct and then compare studies from different populations where the rates of CD without ODD may vary. Finally, the cross-sectional nature of the study may have also limited our ability to detect contextual differences between the ODD + CD and ODD only groups as early manifestations of some of these problems may yield greater differences in adolescence or over time. Longitudinal analyses are needed to test whether these findings can be replicated and persist over time. Larger samples also are needed to study these results in girls.

This study highlights several contextual problems associated with DBDs in clinically referred children, beyond the presence of child dysfunction, which may enhance the trajectory towards antisocial behavior (Lahey et al. 1999). Thus, practitioners and researchers are encouraged to evaluate multiple risk factors related to the parent, family, school, peer, community domains (Burke et al. 2002; Chronis et al. 2003). This expanded assessment scope is especially relevant to young ODD + CD children whose multiple forms of dysfunction and deviance, and exposure to adverse experiences, may require diverse interventions. Our findings also document the striking similarity between children with ODD + CD and those with ODD only in terms of the contextual factors that are often the focus of intervention, despite the consideration of ODD and CD as separate, specific diagnoses.

One clinical implication of these findings is the need to tailor specific interventions in order to address any dysfunctional contextual factors identified in cases presenting with a DBD, particularly poor parental functioning, limited parent management skills, association with deviant peers, and school behavior problems. In fact, almost all of the evidence-based interventions for DBDs focus on changing these contextual factors (see Kolko 2002; Nock 2003). Expanded parental and family interventions seem especially warranted given their documented efficacy and impact on outcomes across both child-specific and general contextual domains, as well as their relevance to the promotion of prosocial peer activities (e.g., Markie-Dadds and Sanders 2006; Shaw et al. 2006). Subsequent analyses from this project will evaluate the role of psychosocial and pharmacologic interventions on the course and outcome of antisocial behavior, and use of multimodal interventions to target clinical problems in multiple domains.