Disruptives Verhalten bei verschiedenen Störungen: Untersuchung einer klinischen Stichprobe mit dem Eyberg Child Behavior Inventory
Abstract
Abstract.Objective: The study reports the prevalence of disruptive behaviors in a help-seeking sample of young children across a diverse range of clinical diagnoses (based on ICD-10). Method: The Eyberg Child Behavior Inventory (ECBI), a parent rating scale of disruptive behaviors, was completed on 310 children (2–11 years) at three child and adolescent psychiatry clinics in three German states (Bavaria, Hesse, Lower Saxony); the majority of children were outpatients. Results: Mean intensity scores of disruptive behaviors differed significantly by diagnostic group, with the lowest ratings within a community sample, and increasingly higher scores in children with a diagnosis from the internalizing spectrum, those with pervasive developmental disorders, and finally, those with externalizing disorders (e. g. hyperkinetic disorder, conduct disorders). Seventy percent of the clinical sample, compared to only 17 % of the community sample, exceeded the normative cut-off score of 111, indicating that disruptive behaviors are common in young German children seeking help for different mental health problems. Conclusions: These findings support the Research Domain Criteria approach by showing that disruptive behaviors cross our current diagnostic labels and may need to be assessed and conceptualized in treatment planning, even in children without a primary diagnosis from the externalizing spectrum.
Zusammenfassung.Zielsetzung: Die Studie berichtet über die Häufigkeit von disruptiven Verhaltensweisen in einer klinischen Stichprobe von jungen Kindern, für die wegen verschiedener klinischer Diagnosen (nach ICD-10) Hilfe gesucht wurde. Methode: An drei kinder- und jugendpsychiatrischen Kliniken in drei Bundesländern (Bayern, Hessen, Niedersachsen) wurde disruptives Verhalten mittels des Eyberg Child Behavior Inventory (ECBI), eines Elternfragebogens, bei 310 Kindern (2–11 Jahre) erfasst; die Mehrzahl der Kinder wurde ambulant behandelt. Ergebnisse: Die mittleren Intensitätswerte für disruptives Verhalten unterschieden sich signifikant in Abhängigkeit von der diagnostischen Gruppe. Der niedrigste Wert ergab sich für eine Stichprobe aus der Allgemeinbevölkerung, gefolgt von Kindern mit einer Störung aus dem internalisierenden Spektrum, Kindern mit tiefgreifenden Entwicklungsstörungen, und den höchsten Werten für Kinder mit externalisierenden Störungen (z. B. hyperkinetische Störung, Störung des Sozialverhaltens). Siebzig Prozent der klinischen Stichprobe, im Vergleich zu 17 % in der Allgemeinstichprobe, überschritten den normativen Cutoff-Wert von 111, was indiziert, dass disruptive Verhaltensweisen bei jungen Kindern in kinder- und jugendpsychiatrischer Versorgung in Deutschland sehr häufig vorkommen. Schlussfolgerungen: Diese Ergebnisse unterstützen den Research Domain Criteria-Ansatz, indem sie zeigen, dass disruptive Verhaltensweisen über unsere gegenwärtigen Diagnosesysteme hinweg auftreten. Sie sollten daher regelmäßig erfasst und in der Therapieplanung berücksichtigt werden, selbst dann, wenn ein Kind keine primäre Diagnose aus dem externalisierenden Spektrum aufweist.
Disruptive behaviors are defined across a wide range of dimensions, including noncompliance, temper tantrums, whining, aggression toward others or lying. Importantly, these behaviors are developmentally expected in early childhood, specifically from age 2 to 7 (Axberg, Hanse, & Broberg, 2008; Wakschlag et al., 2007). However, high levels of disruptive behaviors can be prodromal symptoms of many mental disorders (Burke, Rowe, and Boylanet, 2014; Kim-Cohen et al., 2003; Nock, Kazdin, Hiripi, and Kessler, 2007). Disruptive behaviors are the most common and earliest emerging developmental psychopathologies (Wakschlag et al., 2012), and disruptive disorders like conduct disorder (CD) and oppositional defiant disorder (ODD) are one of the major reasons for referring children to mental health agencies (Steiner, Remsing & The Work Group on Quality Issues, 2007). A recent meta-analysis found that the worldwide prevalence of any disruptive disorder in children and adolescents was 5.7 % (Polanczyk et al., 2015). In preschool children, rates are similar or even slightly higher (Egger & Agold, 2006; Lavigne et al., 2009). For these reasons, it is important to identify children with conduct disorders (ICD-10, F91) early in life. To distinguish “normative” misbehavior (defined as misbehavior occurring in a developmental period where it is expected to occur, Wakschlag et al., 2007) from clinical manifestations of misbehavior is a very important assessment goal and may have significant consequences for treatment recommendations. Therefore, Heinrichs, Bussing, Henrich, Schwarzer, and Briegel (2014) analyzed maternal and paternal reports of child disruptive behavior using the German version of the Eyberg Child Behavior Inventory (ECBI, Eyberg & Pincus, 1999) on more than 5,000 children in pre- and elementary school (ages two to nine) to help discriminate normative from more concerning misbehavior. Based on the Strengths and Difficulties Questionnaire, children were classified as (1) within the normal range, (2) within the threshold range, and (3) within the clinical range of problem behavior. ECBI intensity as well as the problem scores effectively differentiated between these three groups of children. Similar results were found in the USA in non-clinical samples and small clinic-referred samples (for externalizing problems; Burns & Patterson, 1990; Burns, Patterson, Nussbaum & Parkeret, 1991; Eisenstadt, McElreath, Eyberg & McNeil, 1994; Eyberg & Ross, 1978; Eyberg & Robinson, 1983). Similarly, studies from the USA clearly evidenced powerful discrimination between clinic-referred (usually for disruptive behaviors) and non-clinical groups using the ECBI; however, less is known about the extent of disruptive behaviors (assessed with the ECBI) across clinical groups of young children referred for varying types of disorders. Furthermore, clinic-referred samples were usually very small and did therefore not allow for more detailed analyses across different groups of disorders in young children. Weis et al. (2005) included a relatively large clinic-referred sample in their study (n = 115 children, four to six years, maternal reports only). The ECBI intensity score was able to differentiate between three of four groups, assigning the lowest score to the group with no significant externalizing problems (n = 29, M = 90.1, SD = 16.5), followed by children with attention problems (n = 26, M = 126.9, SD = 21.6) and oppositional behavior problems (n = 52, M = 142.3, SD = 30.0) and concluding with the conduct problems group (n = 8, M = 168.9, SD = 24.6). They suggested that future research should use clinical samples for which diagnostic criteria were comprehensively checked and diagnoses were assigned according to classification systems for mental disorders (instead of using a rating scale for one criterion only). Recently, a Dutch study (Abrahamse, Junger, Leijten, Lindeboom, Boer, and Lindauer, 2015) reported on ECBI results for a clinical sample of 197 children aged two to eight years (62 % boys) with significant DSM-IV symptoms. The sample included multi-ethnic groups in an urban region. For 70 % of the sample, diagnostic information was available (n = 137): 51 children did not meet the criteria for a disruptive behavior disorder (ECBI intensity score: M = 111.4, SD = 24.4), 32 were diagnosed with significant ADHD symptoms (ECBI intensity score: M = 134.4, SD = 23.6), 39 with significant ODD symptoms (including children with comorbid significant attention problems; ECBI intensity score: M = 157.4, SD = 28.3) and 14 with CD symptoms (including significant comorbid attention problems and/or oppositional behaviors; ECBI intensity score: M = 162.3, SD = 24.7). No information on other significant comorbid symptoms was given. In contrast to the study of Weis et al. (2005) (and previous studies), which mostly relied on mothers as informants, Abrahamse et al. (2015) also included 79 fathers (40 % of available maternal ratings of the full sample). They reported no significant effect of informant’s gender for the total sample; yet, in a subsample for which both mother and father ratings were available, mothers reported a higher frequency of disruptive behaviors in their children than fathers.
The current study extends previous studies on disruptive behaviors across disorders by examining a much wider spectrum of clinical diagnoses (according to ICD-10, World Health Organization 1992) than in previous studies, using the validated diagnostic status for each participating child and both maternal and paternal ECBI scores. The following three primary aims were defined:
(1) to assess the frequency of disruptive behaviors in young children across different clinical disorders
(2) to evaluate how well the ECBI is able to differentiate between a non-clinical (Heinrichs et al., 2014) and a clinical sample of young children with a variety of mental disorders. We expect the ECBI to successfully differentiate between these two groups with large effect sizes.
(3) to investigate the extent of disruptive behaviors across different groups of ICD-10 disorders. We hypothesized that moderate effect sizes between internalizing and externalizing spectrum disorders would occur when assessed with the ECBI. Finally, we were interested in exploring the extent to which parents perceived the disruptive behaviors as problematic across different clinical diagnoses.
Method
Ethics committees at each study site approved the current study.
Procedure
Participants were recruited through three child and adolescent psychiatry departments in three German cities (Göttingen, Marburg, Schweinfurt) in three states (Lower Saxony, Hesse, Bavaria). While the study center in Schweinfurt is a community mental health clinic, the other departments are university clinics. Recruitment took place from September 2009 to July 2014. At all three sites all families seeking help for a child within the relevant age range (with or without referral) were asked to participate in the present study. Only the study center in Schweinfurt documented acceptance and refusals to allow preliminary analyses on potential differences between participating and refusing families. For consenting families, all sites administered study questionnaires during their routine clinical assessment process. Although the clinical processes at each site were not exactly identical, each site’s routine assessment included a structured and detailed child and family history, psychopathology assessment, physical examination, intelligence testing, administration of disorder-specific questionnaires and interaction analysis in accordance with current German Child and Adolescent Psychiatry Guidelines (Deutsche Gesellschaft für Kinder- und Jugendpsychiatrie und Psychotherapie, Bundesarbeitsgemeinschaft Leitender Klinikärzte für Kinder- und Jugendpsychiatrie und Psychotherapie & Berufsverband der Ärzte für Kinder- und Jugendpsychiatrie und Psychotherapie, 2007). If necessary, neuropsychological assessment (e. g. regarding attention), assessment by a speech, occupational or movement therapy and further somatic measures, such as blood tests, ECG and EEG, were added. Upon completion of patients’ assessments, a multi-disciplinary consensus process led by board certified child and adolescent psychiatrists was used to assign diagnoses according to the multiaxial classification of child and adolescent psychiatric disorders of the World Health Organization (MAS; World Health Organization 1992; German version Remschmidt, Schmidt & Poustka, 2012) and the German child and adolescent psychiatry guidelines (Deutsche Gesellschaft für Kinder- und Jugendpsychiatrie und Psychotherapie et al. 2007). Clinicians were not allowed to use study-derived information from the ECBI during the diagnostic process.
Materials
Sociodemographic Characteristics
The study questionnaire included 33 items assessing sociodemographic characteristics as well as medication and chronic physical and/or mental diseases diagnosed prior to the assessment visit at the respective study site. Information was obtained about children (age and gender) and each parent (e. g. age, gender, country of birth and educational level). Twenty-six of the questionnaire items were taken from the KIGGS survey (Ravens-Sieberer, Wille, Bettge, and Erhart, 2007). Socioeconomic status (SES) was determined accordingly (Lange, Kamtsiuris, Lange, Schaffrath, Stolzenberg, and Lampert, 2007).
Eyberg Child Behavior Inventory (ECBI)
The ECBI consists of 36 items describing disruptive behaviors. Each item rating (on a scale from 1 to 7) reflects the parental perception of behavior frequency (summed to obtain the intensity scale; ranging from 36–252) and whether this behavior is considered problematic or not (problem scale; ranging from 0–36). The problem scale is also considered an indicator of parent stress caused by the child’s behavior. Psychometric results of the German version of the ECBI derived from a community-based sample (Heinrichs et al., 2014) showed high internal consistency for both scales on each age level (two to nine years; Cronbach’s alpha: between .89 and .94 for each age group and for both mother and father ratings). Means of both the intensity scale and the problem scale were found to be significantly lower than in the USA, resulting in recommended clinical problem behavior cut-off scores of 111 (intensity scale) and 12 (problem scale; Heinrichs et al., 2014) (compared to 132 and 15 in the USA, respectively, Colvin Eyberg, and Adams, 1999). There is considerable evidence that the German ECBI is a reliable and valid instrument to screen for significant disruptive behaviors in early and middle childhood in a non-clinical sample. The correlation between maternal and paternal report was r = 0.77 for the intensity scale and r = 0.62 and r = 0.57 (non-parametric) for the problem scale in the present sample.
Strengths and Difficulties Questionnaire (SDQ)
The SDQ (Goodman, 1997) consists of 25 items describing child behaviors and traits. Each item is rated either by parent, teacher or the (older) child according to how typical the psychological attribute is for the child (0 = not true, 1 = somewhat true, 2 = certainly true). The SDQ is comprised of five scales (each scale includes five items) and the total difficulties score (sum of the four difficulty scales, score ranging from 0 to 40). In the present study, only the total difficulties score was used, which is usually the most reliable score (e. g. Lohbeck, Schultheiß, Petermann, and Petermann, 2015). The SDQ was used because it is a validated instrument in Germany and was shown to differentiate between clinical and non-clinical samples. The correlation between maternal and paternal report was r = 0.77 in the present sample.
Participants
Parents of 310 children (two to 11 years; mean age in months: M = 72.9, SD = 19.6) provided informed consents and participated in the current study. The vast majority of children (n = 281, 91 %) were outpatients (day-clinic patients: n = 10, inpatients: n = 17; no data provided: n = 2). Mean ages of mothers and fathers were 35.8 (SD = 6.1) and 39.4 (SD = 7.2) years (with n = 3 and n = 43 no data provided in the available sociodemographic information, respectively). Eighty percent of children were male (n = 248). The majority of children lived with their biological parents (63 %), 22 % in single parents households (including five single-parent fathers and 59 single-parent mothers) and 10 % with one biological parent and his/her new partner. The remaining children were adopted or in foster care. Most children had siblings (n = 240, 77 %). Seventeen percent of parents (n = 54) did not complete all necessary items to construct the SES level (e. g. income). Of the remaining families, 27 % were classified with low, 36 % with middle and 20 % with high SES. Seven percent of families had a migration background (i. e. according to the KIGGS definition both parents must have a migration background). Two hundred and sixty-three children had been diagnosed with a mental disorder on the first axis according to ICD-10, while 43 children presented with problems which were either not severe enough or did not qualify for a diagnosis for other reasons. Diagnostic status was missing for four children.
Approximately one-third of the sample (31 %) had at least one additional diagnosis assigned. Comorbidity was distributed across disorders: 24 % had an additional diagnosis from the internalizing spectrum and 29 % from a spectrum other than internalizing or externalizing, with the remaining 47 % having comorbid disorders from the externalizing spectrum (27 % hyperkinetic disorder, 16 % conduct disorder including oppositional defiant disorder and 4 % hyperkinetic conduct disorder). Table 1 illustrates the sample based on the MAS.
Classification of Groups of Clinical Disorders
In the first step, the main axis I diagnosis according to ICD-10 was used to group children into one of eight categories: (0) symptoms but no mental disorder (n = 43), (1) attachment disorders (F94.1 and F94.2; n = 6), (2) emotional disorders (F32, F4, F93 and F94.0; n = 43), (3) pervasive developmental disorder (F84; n = 41), (4) hyperkinetic disorder, disturbance of activity and attention (F90.0; n = 67), (5) hyperkinetic conduct disorder (F90.1; n = 37), (6) conduct disorder (F91; n = 35), (7) other disorders, residual category (F50–52, F 92.9, F95–F98.9 including 12 children with F98.8; n = 31). In the second step, two broader groups were defined: (1) externalizing disorders (F90–F91; n = 139), and (2) internalizing disorders (F32, F4, F93 and F94.0; n = 43). Finally, because some of the diagnoses on axis I were considered “suspected” (indicating that a clear diagnostic assignment was not possible based on the initial assessment), we repeated the primary analyses excluding children with suspected diagnoses (n = 108). The results were similar, hence, we will only report the results for the whole sample here.
Data Analysis
All statistical analyses were performed using the software package SPSS.
Missing Data
If more than four items on the ECBI intensity or problem scale were missing, the respective scale was not included in further analyses, which resulted in the exclusion of 17 (6 %) maternal and 91 (29 %) paternal intensity scale ratings, and of 29 (9 %) maternal and 97 (31 %) paternal problem scale ratings (including single-parent household where the secondary caregiver did not complete questionnaires at all). If four or less items were not completed, the mean of the respective scale replaced the missing value. This approach resulted in a total of 293 maternal and 219 paternal ECBI intensity scores and 281 and 213 ECBI problem score ratings, respectively.
SDQ scales for which more than two items per scale were not completed, were excluded from further analyses. For up to two missing items per scale, the mean scale score was used to replace the missing value. The total score represents the sum of the four symptom scales.
Statistical Procedures
To examine differences between groups, we either conducted univariate analysis of variance (including post-hoc Tamhane tests to follow-up on overall significant group effects with the diagnostic group as the independent variable and the maternal or paternal intensity score as the dependent variable) or Kruskal-Wallis tests (dependent on distribution of data) and t-tests (or Mann-Whitney U tests). In order to continuously use the largest sample available, separate analyses for mothers and fathers, and for intensity and problem scale scores, were conducted. Finally, we also tested if informant discrepancies occurred if the disorder group was included in the analyses (with a repeated measure analysis of variance due to the dependency of mother and father ratings). Cohen’s d was calculated for differences of group means using pooled SD and correcting for the different sample sizes (ghedges).
Results
Study participants vs. refusers
No differences were found when comparing n = 48 refusing to n = 119 participating families from the Schweinfurt site with regard to child age and gender, treatment setting, medication (yes/no), main axis one ICD-10 diagnosis (classified into the above described groups), intelligence level (below 85, 85–115, above 115), somatic disorders (yes/no), current abnormal psychosocial circumstances (yes/no), or psychosocial functioning (all based on the multi-axial system of the ICD-10).
Mean Scores for ECBI Intensity and Problem Scores Across Disorders
Mean scores for maternal and paternal ECBI intensity and problem ratings as well as SDQ total scores for all eight diagnostic groups are illustrated in Table 2. The total mean intensity score for the clinical sample was 125.3 and 125.1 (SDmothers = 32.8, SDfathers = 30.9), with mother ECBI scores ranging from 45 to 221 and father ECBI scores ranging from 55 to 235. Mean scores for maternal and paternal problem ratings were 14.4 (SD = 7.9) and 13.1 (SD = 8.0), respectively, and scores ranged from 0 to 33 for mothers and from 0 to 35 for fathers. Mean scores for maternal and paternal SDQ total scores were 17.0 (SD = 6.3) and 16.3 (SD = 6.3), respectively, and scores ranged from 2 to 34 for mothers and from 2 to 33 for fathers.
Associations of ECBI Scores with Sociodemographic Variables
Maternal and paternal ECBI intensity scale scores, but not maternal problem scale scores, were normally distributed. We therefore only tested the ECBI intensity scale score for age, gender or informant (maternal/paternal rating) effects, conducting several univariate analyses.
Effects of Child Age
No significant effects of child age on maternal or paternal ECBI intensity scores were found (seven age cohorts: 2–3 years, 4 years, 5 years, 6 years, 7 years, 8 years, 9–11 years), F(6, 286) = 1.9, p = .080, η2 = .039 for mothers, and F(6, 212) = 0.77, p = .600, η 2 = .021 for fathers. The explained variance was below 4 %.
Effects of Child Gender
There was a significant effect of child gender on maternal and paternal ECBI intensity scores, F(1, 291) = 8.4, p = .004, η 2 = .028 for mothers, and F(1, 217) = 4.3, p = .039, η 2 = .019 for fathers. Mothers’ mean scores for sons and daughters were 128.1 (SD = 31.5) and 114.2 (SD = 36.5), respectively. Fathers’ mean scores for sons and daughters were 127.3 (SD = 30.1) and 116.1 (SD = 32.8), respectively. These results indicate that boys usually scored slightly higher than girls on the ECBI intensity scale. However, the explained variance was below 3 % for both informants.
Effects of Informant
Only children for whom both mother and father ratings were available were included in this analysis (n = 208). There was no significant difference regarding intensity scores between mothers and fathers, t(1, 207) = 0.91, p = .360. When including the specific group of disorders, we also did not find a significant main effect of informant (F(1,198) = 1.3, p = 0.25) or an interaction between informant and disorder group across the eight groups on the ECBI intensity scale score, F(7,198) = 0.38, p = 0.91. However, we did find a significant main effect of informant on the ECBI problem scale score (F(1,191) = 5.4, p = 0.02) but no interaction between informant and disorder group (F(7,191) = 0.69, p = 0.69), indicating that fathers perceive ECBI-related behavior problems as less problematic than mothers (with both informants rating similar mean levels of behavior frequency). When repeating these analyses for the well validated SDQ total score, we also failed to find a statistical significant interaction effect, but did find a significant main effect of informant (F(1,200) = 8.3, p = 0.004) indicating lower SDQ total scores in fathers than mothers (averaged across disorder groups).
Effects of Socioeconomic Status
There was a significant effect of socioeconomic status (low, middle, high) on intensity ratings of mother ratings (F(2, 241) = 4.1, p = .018) and a non-significant effect on father ratings, F(2, 188) = 3.0, p = .055. Mothers’ mean scores for low to high SES were 132.6 (SD = 35.2), 125.1 (SD = 32.3) and 116.3 (SD = 28.9), indicating a decrease with increasing SES. Similarly, fathers’ mean scores decreased from 133.6 (SD = 33.8) to 125.7 (SD = 33.0) and to 118.1 (SD = 24.9) with increasing SES. Only the post-hoc comparison of low versus high SES reached statistical significance with effect sizes between dMothers = 0.41 and dFathers = 0.53.
Aim 1: Discrimination between children with and without a mental disorder
As expected, a two-tailed t-test for independent samples indicated a statistically significant difference between the community sample (Heinrichs et al. 2014, MMothers = 87.8, MFathers= 88.8) and the present sample of children with mental disorders for both maternal (M = 127.1) and paternal (M = 126.9) intensity scores, t = 25.6, df = 5270, p < .001, d = 1.67, and t = 21.6, df = 4119, p < .001, d = 1.63. The group of children with emotional disorders demonstrated lower mean scores on the ECBI than the total clinical sample; therefore, we compared this specific group to the community sample. We found a statistically significant difference between both maternal (M = 103.4) and paternal (M = 112.6) intensity scores compared to the community sample, t = 4.3, df = 5064, p < .001, d = 0.68, and t = 5.5, df = 3963, p < .001, d = 1.04.
Aim 2: Extent of disruptive behaviors across diagnostic groups
Means, standard deviations and sample size per diagnostic group are illustrated in Table 3.
Differentiation between externalizing and internalizing disorder groups
As expected, a two-tailed t-test for independent samples indicated a statistically significant difference for maternal (m) and paternal (p) intensity scores, tm = 6.5, df = 171, p < .001, dm = 1.16, and tp = 3.7, df = 118, p < .001, dp = 0.80. Problem scores were also significantly different, Zm = 5.5, p < .001, d = 1.1 and Zp = 2.6, p = .010, d = 0.56, respectively. The same result was found, when we excluded cases with comorbid disorders of the externalizing spectrum from the internalizing group and vice versa: maternal and paternal reports on the ECBI intensity scale for internalizing disorders (MM = 99.1, SDM = 28.6; MF = 110.7, SDF = 27.6) and externalizing disorders (MM = 139.0, SDM = 31.0; MF = 139.2, SDF = 31.8) were still statistically different (for mothers: t = 6.79, df = 148, p < 0.001; for fathers: t = 3.89, df = 100, p < 0.001). Similarly, problem scores were significantly different: ZM = 5.6, p < .001, and ZF = 2.3, p = .020.
Differentiation between Pervasive Developmental Disorders and Externalizing Disorders
A two-tailed t-test for independent samples indicated a statistically significant difference for maternal and paternal intensity scores comparing pervasive developmental disorders to disorders from the externalizing spectrum, tm = 3.5, df = 168, p = .001, d = 0.64, and tp = 2.6, df = 123, p = .011, d = 0.52. Problem scores were also significantly different, Um = 3.8, p < .001, d = 0.76 and Up = 3.7, p < .001, d = 0.75, respectively.
Differentiation between Disturbance of Activity and Attention (F90.0), Hyperkinetic Conduct Disorder (F90.1) and Conduct Disorders (F91)
A significant group effect was found in a univariate analysis of variance with F(2,130) = 13.2, p < .001, η 2 = .17 and F(2,89) = 10.1, p < .001, η 2 = .19 for mother and father intensity scores, respectively. Post-hoc Tamhane tests for mothers demonstrated significantly different scores between disturbance of activity and attention (F90.0) and conduct disorders (F91), mdiff = 16.1, p = .040, d = 0.56 as well as hyperkinetic conduct disorder (F90.1), mdiff = 30.4, p < .001, d = 1.09. However, the conduct disorders group was not significantly different from the hyperkinetic conduct disorder group (F90.1), mdiff = 14.3, p = .142, d = 0.48. Post-hoc Tamhane tests for fathers demonstrated no significant difference between disturbance of activity and attention (F90.0) and conduct disorders (F91), mdiff = 3.3, p = .960, d = 0.13, but yielded a significant difference between these disorders and hyperkinetic conduct disorder (F90.1), mdiff = 32.9, p = .003, d = 1.16 and mdiff = 29.6, p = .021, d = 0.85. Similarly, a Kruskal-Wallis test yielded significant differences across problem scores for both mothers and fathers (χ2 = 15.0, df = 2, p = .001; χ 2 = 8.8, df = 2, p = .012, respectively). Follow-up tests were only conducted for mothers because the paternal problem score means for disturbance of activity and attention (F90.0) and conduct disorders (F91) were similar. Mann-Whitney U tests indicated no significant differences for maternal problem scores between F90.0 and F91 (d = 0.35) but did between F91 and F90.1 (d = 0.54) as well as between F90.0 and F90.1 (d = 0.90). In sum, maternal ratings seem to differentiate better between hyperkinetic and conduct disorders, while paternal ratings discriminated better between hyperkinetic conduct disorder and conduct disorders.
Discussion
The present study aimed at using a large clinical sample to identify the extent of disruptive behaviors occurring across groups of ICD-10 diagnoses, including hyperkinetic (F90.0), conduct disorder (F91) with mostly oppositional-defiant disorder (F91.3) and their combination (F90.1). We employed the ECBI and included not only maternal, but also paternal estimates for these behaviors. As a main result, we found that 70 % of the clinical sample exceeded the community-derived cut-off score (compared to 17 % in the community sample), indicating that disruptive behaviors are a common problem in young children in Germany seeking help for different mental health problems and that disruptive behavior is not limited to disorders from the externalizing spectrum.
In preliminary analyses, we did not find an indication for selective participation in the study. However, this was only documented in one of three recruitment sites and does therefore not exclude sample selection effects. We also checked for age, gender and informant effects and found similar results as in previous studies, with boys yielding higher scores than girls (also reported for example in Abrahamse et al., 2015 or Reedtz et al., 2008). Across studies and samples, small effect sizes were reported for this gender difference, which is also in line with our result. Similarly, all current effects on other sociodemographic variables (e. g. SES or age) were similar to those reported in the literature. Although statistical significance may be different across studies (e. g. there was no statistical difference for age in the present sample), effect sizes in the proportion of explained variance are relatively consistent: usually only small effects occur (e. g. Burns & Patterson, 1990; Colvin et al., 1999) and they may be less meaningful for clinical than for research purposes. An exception might be the association between SES and level of parent-reported disruptive behavior, which seems higher than for other sociodemographic indicators. It is unclear if children from low SES families indeed show more disruptive behaviors, or if parents attribute more child behaviors to the disruptive category compared to high SES families.
As expected, ECBI intensity as well as problem scale scores easily discriminated a clinical sample from a community sample with very large effect sizes. The same applies to the SDQ. Notably, the community sample is not interchangeable with the “non-clinical” group in this study: we also had 43 children who presented in one of the outpatient services but ended up not meeting criteria for any ICD-10 diagnoses. This group displayed considerable disruptive behavior levels (Table 2) and is quite similar to the clinic-referred subsample in the Dutch study (Abrahamse et al., 2015), where children did not meet threshold for a DSM-IV diagnosis. ECBI intensity as well as problem scores were very similar in both samples, indicating that disruptive behaviors – although not fulfilling diagnostic criteria – are a substantial burden for the families and lead to the referral of children to mental health agencies (Steiner et al., 2007). While both ECBI and SDQ scores could discriminate a subthreshold symptom group from a clinical group consisting of different ICD-10 diagnoses, the ECBI appears to do so more successfully than the SDQ, when considering the number of significant post-hoc tests. However, due to the small sample sizes of some disorder groups, this conclusion needs to be considered with caution.
High ECBI intensity scale scores were found not only for children with conduct disorders (F91, F90.19), but also for children with attachment disorders, disturbance of activity and attention as well as pervasive developmental disorder and even the “symptoms only” group, which exhibits mean intensity scores above the normative German cut-off of 111. Furthermore, children with emotional disorder of childhood score higher than this normative mean on the father scores and display a very large standard deviation in the evaluation of mothers.
When comparing ICD-10 diagnoses from the internalizing spectrum with those from the externalizing spectrum, ECBI intensity scores differ by group with large effect sizes. Children with a diagnosis from the internalizing spectrum display higher mean ECBI scores than those from a community sample; however, they are still approximately one standard deviation below children of the other clinical groups. Interestingly, the diagnostic group of “emotional disorders” seems to be the only one in which fathers rate symptoms to be somewhat more frequent than mothers do. This is against the common trend of informant discrepancies in which fathers on average show a lower mean score than mothers across disorder groups. Future studies may specifically explore this pattern to determine if this is systematic or accidental.
Weis and colleagues (2005) reported a mean ECBI intensity score of 90.1 (SD = 16.5) for children referred to outpatient psychology services, with no external behavior problem presentation (Weis et al., 2005), a score clearly lower than the one we found in children with a primary disorder from the internalizing spectrum. One explanation may be that the children in the Weis et al. study (2005) did not have any significant problems from either the externalizing or internalizing spectrum, and therefore constituted a sample more similar to a community sample. Another explanation may be that children with internalizing problems in our study also showed, to some extent, behavioral problems. Thus, the higher mean score in our sample may be slightly biased by other, secondary diagnoses present. However, additional analyses, excluding those cases of cross-spectrum comorbidity, demonstrated still a very similar mean score. Thus, the result and conclusion that emotional disorders are also characterized by increased levels of disruptive behaviors were not impacted or challenged by this type of comorbidity. Also, fathers seem to notice more disruptive behaviors in children with internalizing disorders than mothers, although they report similar levels of problem perception (see Table 2).
Furthermore, we assumed that successful discrimination would also be possible when comparing diagnoses with different levels of (significant) behavior problems, such as pervasive developmental disorder. Like several population-based studies (Levy et al., 2010; Simonoff et al., 2008; Totsika, Hastings, Emerson, Lancaster, and Berridge, 2011), we also found high levels of disruptive behavior in children with pervasive developmental disorder. One study (Stuttard et al., 2014), using the ECBI in a sample of children with autism and intellectual disability, found even higher scores than in our study. These results reveal that disruptive behavior is not a specific symptom indicative of conduct disorder/hyperkinetic conduct disorder. Even though the group of externalizing disorder holds significantly higher scores than the internalizing and pervasive developmental disorder group, disruptive symptoms also occur to a clinically relevant degree in the latter groups. The limited research available on comorbidity in preschool children suggests symptom co-occurrence between the broad categories of emotional and behavioral disorders (Egger & Angold, 2006; Franz, Angold, Copeland, Costello, Towe-Goodman, and Egger, 2013). One may criticize that the present findings are not surprising, as the ECBI items are in part similarly worded to those of ADHD and ODD symptomology and few are related to internalizing symptoms. However, it is then specifically surprising to (also) find the relatively high prevalence of disruptive behavior problems in children with emotional disorders. This study adds to this emerging literature in important ways by examining disruptive behavior across various clinical disorders, including emotional disorder.
Finding a high frequency of disruptive behaviors in young children across different disorders could also be interpreted as ambiguous boundaries with considerable overlap between mental disorders. Categorical diagnoses have “clinical utility, but they necessarily reduce the complexity and heterogeneity of clinical phenomenology” (Wakschlag et al., 2012, p. 593). Our results support the concept of complex clinical phenotypes as clinical profiles of symptoms rather than distinct categories to enable the identification of targeted intervention.
In comparison, the SDQ does not seem to differentiate well between some diagnostic groups, e. g. the “symptoms only group” and the “emotional disorders group”. Furthermore, the instrument revealed similar scores for disturbances of activity and attention and conduct disorder. On the other hand, the SDQ – like the ECBI – picked up well the difference of attachment disorders to many other clinical groups and it also peaked when parents of children with hyperkinetic conduct disorders completed the questionnaire. Finally, the similarity in result pattern on informant discrepancies between the SDQ total score and the ECBI problem scale score hint to the possibility that in SDQ symptom ratings both frequency and problem perception are intermingled.
Obviously, ECBI scores corresponded nicely to increasing severity of disruptive symptom presentation as assessed by clinicians. This is in line with previous studies reporting that childhood aggression differs in quantity (along a dimension of severity) rather than in quality or distinct categories (Walters, Ronen, and Rosenbaum, 2010). It has been stated before that attention problems have the strongest association with conduct problems, even in non-clinical samples (Kirkhaug, Drugli, Lydersen, and Mørch, 2013). Thus, the difficulty of discriminating between hyperkinetic and oppositional-defiant disorder may just reflect this strong association. In a recent study, deVries and colleagues (2016) found parents of six to 12 year old children to be more likely to “(a) rate a child as inattentive in the presence of hyperactivity symptoms, (b) rate a child as oppositional in the presence of inattention and hyperactivity symptoms, and (c) more likely to rate a child as inattentive and hyperactive in the presence of oppositional symptoms”. However, these negative halo effects were less pronounced in parents than in teachers (or college students) and parents may therefore be the best choice for rating child disruptive behavior.
The present study has a number of limitations that need to be carefully considered when interpreting results. First, research was implemented during routine clinical assessments and only the study center in Schweinfurt systematically documented participation and refusal rates. Furthermore, findings reflect the recruitment through child psychiatric hospitals and the resulting prevalence rates of included disorders. Thus, the validity of the present results is dependent on the validity of the clinical diagnosis and the associated assessment procedures. Also, in this routine clinical care sample, one third of the sample had more than one disorder, complicating the interpretation of results because the non-primary diagnosis may contribute to the extent of externalizing problem behaviors. However, at least for the comparison of internalizing and externalizing groups this effect was rather small. Yet, clinicians noted that they were unsure if one of the disorders from the externalizing spectrum (F90.0 and F91) truly occurred isolated or if there were additional, significant symptoms that would meet threshold for assigning the combination disorder (F90.1). In general, comorbidity was higher within a spectrum than across spectrum. This again emphasizes the strong association between this symptom complex. Another limitation may be that the present study included relatively few girls, although this may be in line with gender distributions for respective disorders. Abrahamse and colleagues (2015) also reported 62 % boys in their clinical sample. Therefore, generalization of our findings to girls is limited. Furthermore, maternal ECBI problem scores were not normally distributed, limiting several comparisons by parental gender. And finally, we did not assess parental mental health, precluding the ability to control for this potentially confounding variable. Parental mental disorder may result in biased perceptions of the child behavior; this is a general methodological challenge when assessing child behavior via parent report.
Results from the ECBI problem scale score were used to assess parental stress. It would have been useful to employ an actual parenting stress measure; however, the extent of possible assessment routines in clinical care is limited and a number of studies on the ECBI validated this inference. Finally, our results relate to the ICD-10 classification system. According to the ICD-10, hyperkinetic disorder and oppositional-defiant disorder (as a specified subtype of conduct disorders) are viewed more closely related than in the latest version of the DSM (DSM-5), where ADHD is assigned to the category of neurodevelopmental disorders. This difference in classification needs to be addressed when interpreting our results.
We suggest that our findings have important implications for the development of individualized assessment and treatment planning: disruptive behavior seems to be a prevalent comorbid symptom across multiple disorders and this suggests that the evaluation of symptom profiles is equally important as categorical diagnostic evaluation. The results underline the urgent need for appropriate assessment tools – including the ECBI – to evaluate disruptive behaviors in different diagnostic groups. They also point at the importance of targeting disruptive behaviors in intervention plans, even in disorders from the internalizing spectrum, which usually show low categorical comorbidity with disorders from the externalizing spectrum. Future research may consider including externalizing behavior as a (secondary) outcome domain for children with internalizing disorders.
Conflicts of interests: Regina Bussing serves as International Development Task Force chairwoman of PCIT International, Inc. Wolfgang Briegel is a PCIT Master Trainer and serves as International Development Task Force member of PCIT International, Inc.
Literature
(2015). Psychometric Properties of the Dutch Eyberg Child Behavior Inventory (ECBI) in a Community Sample and a Multi-Ethnic Clinical Sample. J Psychopathol Behav, 37(4):679–691. doi: 10.1007/s10862-015-9482-1
(2008). Parents’ description of conduct problems in their children – A test of the Eyberg Child Behavior Inventory (ECBI) in a Swedish sample aged 3–10. Scandinavian Journal of Psychology, 49:497–505. doi: 10.1111/j.1467-9450.2008.00670.x
(1997). Defiant children. New York: Guilford Press.
(2014). Disruptive Behavior Disorders: Taking an RDoC(ish) Approach. Current Topics Behav Neurosci, 16: 319–336. doi: 10.1007/7854_2013_247
(2014). Functional outcomes of child and adolescent oppositional defiant disorder symptoms in young adult men. Journal of Child Psychology and Psychiatry, 55(3), 264–272. doi: 10.1111/jcpp.12150
(1990). Conduct Problems Behaviors in a Stratified Random Sample of Children and Adolescents: New Standardization Data on the Eyberg Child Behavior Inventory. Psychological Asssessment, 2(4), 391–397. doi: 10.1037/1040-3590.2.4.391
(1991). Disruptive Behaviors in an Outpatient Pediatric Population: Additional Standardization Data on the Eyberg Child Behavior Inventory. Psychologial Assessment, 3(2), 202–207. doi: 10.1037/1040-3590.3.2.202
(1999). Restandardization of the Eyberg Child Behavior Inventory. http://pcit.phhp.ufl.edu/Measures/ECBI%20Restandardization.pdf. Accessed 28 June, 2015.
Conduct Problems Prevention Research Group (2011). The Effects of the Fast Track Preventive Intervention on the Development of Conduct Disorder Across Childhood. Child Development, 82(1), 331–345. doi: 10.1111/j.1467-8624.2010.01558.xDeutsche Gesellschaft für Kinder- und Jugendpsychiatrie und Psychotherapie, Bundesarbeitsgemeinschaft Leitender Klinikärzte für Kinder- und Jugendpsychiatrie und Psychotherapie ,Berufsverband der Ärzte für Kinder- und Jugendpsychiatrie und Psychotherapie (2007). Leitlinien zur Diagnostik und Therapie von psychischen Störungen im Säuglings-, Kindes- und Jugendalter, 3rd edn. Köln.(2016). Negative Halo Effects in Parent Ratings of ADHD and ODD. Psychopathology and Behavioral Assessment. doi:10.1007/s10862-016-9560-z
(2006). Common emotional and behavioral disorders in preschool children: presentation, nosology, and epidemiology. J Child Psychol Psyc, 47(3–4), 313–337. doi: 10.1111/j.1469-7610.2006.01618.x
(1994). Interparent Agreement on the Eyberg Child Behavior Inventory. Child Fam Behav Ther, 16(1), 21–27. doi: 10.1300/J019v16n01_02
(1999). Eyberg Child Behaviour Inventory and Sutter-Eyberg Student Behavior Inventory – Revised: Professional Manual. Odessa, Florida.
(1983). Conduct Problem Behavior: Standardization of a Behavioral Rating Scale with Adolescents. J Clin Child Psychol, 12(3), 347–354. doi: 10.1080/15374418309533155
(1978). Assessment of child behavior problems: The validation of a new inventory. J Clin Child Psychol, 7(2), 113–116. doi: 10.1080/15374417809532835
(2013). Preschool Anxiety Disorders in Pediatric Primary Care: Prevalence and Comorbidity. J Am Acad Child Psy, 52(12), 1294–1303. doi: 10.1016/j.jaac.2013.09.008
(1997). The Strengths and Difficulties Questionnaire: A Research note. J Child Psychol Psyc, 38(5), 581–586. doi: 10.1111/j.1469-7610.1997.tb01545.x
(2014). Die Erfassung expansiven Verhaltens mit der deutschen Version des Eyberg Child Behavior Inventory (ECBI) [The assessment of child behavior problems with the German version of the Eyberg Child Behavior Inventory (ECBI)]. Diagnostica, 60:197–210. doi: 10.1026/0012-1924/a000115
(2003). Prior Juvenile Diagnoses in Adults With Mental Disorder: Developmental Follow-Back of a Prospective-Longitudinal Cohort. Arch Gen Psychiat, 60(7), 709–717. doi: 10.1001/archpsyc.60.7.709
(2013). Associations between high levels of conduct problems and co-occurring problems among the youngest boys and girls in schools: A cross-sectional study. Nord J Psychiat, 67(4), 225–232. doi: 10.3109/08039488.2012.731432
(2007). Messung soziodemographischer Merkmale im Kinder- und Jugendgesundheitssurvey (KiGGS) und ihre Bedeutung am Beispiel der Einschätzung des allgemeinen Gesundheitszustandes [Sociodemographic characteristics in the German Health Interview and Examination Survey for Children and Adolescents (KiGGS) – operationalisation and public health significance, taking as an example the assessment of general state of health]. Bundesgesundheitsblatt, 50(5), 578–589. doi: 10.1007/s00103-007-0219-S
(2009). The prevalence of ADHD, ODD, Depression, and Anxiety in a Community Sample of 4-Year-Olds. J Child Psychol Psyc, 38(3), 315–328. doi: 10.1080/15374410902851382
(2015). Die deutsche Selbstbeurteilungsversion des Strengths and Difficulties Questionnaire (SDQ-Deu-S). Psychometrische Eigenschaften, Faktorenstruktur und Grenzwerte. Diagnostica, 61, 222–235.
(2010). Autism Spectrum Disorder and Co-occurring Developmental, Psychiatric, and Medical Conditions Among Children in Multiple Populations of the United States. J Dev Behav Pediatr, 31(4), 267–275. doi: 10.1097/DBP.0b013e3181d5d03b
(2007). Lifetime prevalence, correlates, and persistence of oppositional defiant disorder: results from the National Comorbidity Survey Replication. J Child Psychol Psyc, 48(7), 703–713. doi: 10.1111/j.1469-7610.2007.01733.x
(2014). Testing Models of Psychopathology in Preschool-aged Children Using a Structured Interview-based Assessment. J Abnorm Psychol, 42(7), 1201–1211. doi: 10.1007/s10802-014-9865-x
(2015). Gene-environment interactions in psychopathology throughout early childhood: a systematic review. Psychiat Genet, 25(6), 223–233. doi: 10.1097/YPG.0000000000000106
(2015). Annual Research Review: A meta-analysis of the worldwide prevalence of mental disorders in children and adolescents. J Child Psychol Psyc, 56(3), 345–365. doi: 10.1111/jcpp.12381
(2007). Psychische Gesundheit von Kindern und Jugendlichen in Deutschland. Ergebnisse aus der BELLA-Studie im Kinder- und Jugendgesundheitssurvey (KiGGS) [Mental health of children and adolescents in Germany. Results from the BELLA study within the German Health Interview and Examination Survey for Children and Adolescents (KiGGS)]. Bundesgesundheitsblatt, 50(5), 871–878. doi: 10.1007/s00103-007-0250-6
(2008). Eyberg Child Behavior Inventory (ECBI): Norwegian norms to identify conduct problems in children. Scand J Psychol, 49(1), 31–38. doi: 10.1111/j.1467-9450.2007.00621.x
(2012). Multiaxiales Klassifikationsschema für psychische Störungen des Kindes- und Jugendalters nach ICD-10 der WHO: mit einem synoptischen Vergleich von ICD-10 und DSM-IV, 6th edn. Bern.
(2008). Psychiatric Disorders in Children With Autism Spectrum Disorders: Prevalence, Comorbidity, and Associated Factors in a Population-Derived Sample. J Am Acad Child Psy, 47(8), 921–929. doi: 10.1097/CHI.0b013e318179964f
(2007). Practice Parameter for the Assessment and Treatment of Children and Adolescents With Oppositional Defiant Disorder. J Child Psychol Psyc, 46(1), 126–141. doi: 10.1097/01.chi.0000246060.62706.af
(2014). Riding the Rapids: Living with autism or disability – An evaluation of a parenting support intervention for parents of disabled children. Res Dev Disabil, 35(10), 2371–2383. doi: 10.1016/j.ridd.2014.05.021
(2011). A population-based investigation of behavioural and emotional problems and maternal mental health: associations with autism spectrum disorder and intellectual disability. J Child Psychol Psyc, 52(1), 91–99. doi: 10.1111/j.1469-7610.2010.02295.x
(2007). A developmental framework for distinguishing disruptive behavior from normative misbehavior in preschool children. J Child Psychol Psyc, 48(10), 976–987. doi: 10.1111/j.1469–7610.2007.01786.x
(2014). Advancing a Multidimensional, Developmental Spectrum Approach to Preschool Disruptive Behavior. J Am Acad Child Adolesc Psychiatry, 53(1), 82–96.e3. doi:10.1016/j.jaac.2013.10.011
(2012). Putting Theory to the Test: Modeling a Multidimensional, Developmentally-Based Approach to Preschool Disruptive Behavior. J Child Psychol Psyc, 51(6), 593–604. doi: 10.1016/j.jaac.2012.03.005
(2010). The latent structure of childhood aggression: A taxometric analysis of self-reported and teacher-rated aggression in Israeli schoolchildren. Psychol Assessment, 22(3), 628–637. doi: 10.1037/a0019779
(2005). Factor Structure and Discriminative Validity of the Eyberg Child Behavior Inventory With Young Children. J Psychopathol Behav, 27(4), 269–278. doi: 10.1007/s10862-005-2407-7
World Health Organization (1992). The ICD-10 Classification of Mental and Behavioural Disorders. Clinical descriptions and diagnostic guidelines. Geneva.
