Background
A critical issue in the field of psychiatry is how to best describe and understand the nature of impulsive aggression in children and adolescents. It is well documented that individuals with early-onset disruptive behavior, including attention-deficit/hyperactive (ADHD), conduct (CD), and oppositional defiant (ODD), disorders are at high risk for later adverse psychosocial outcomes (e.g., school dropout, criminality, substance abuse, reduced social skills, and mental health problems) [
1]. Not surprisingly, disruptive behavior disorders (DBD) are common in early life and have reported lifetime rates of 8.1% for ADHD, 12.6% for ODD, and 6.8% for CD [
2]. While much less studied in children and adolescents, Intermittent Explosive Disorder (IED), a disorder of recurrent, problematic, impulsive aggression, is also common in young individuals and has a reported lifetime prevalence of 7.8% [
3]. Despite the fact that these four disorders share overlapping behaviors, there is no published data examining the relationship between the DBDs and IED.
The validity of IED in adults is now supported by studies showing that IED: (a) can be diagnosed reliably [
4], (b) is relatively stable over time [
5], (c) is taxonic rather than dimensional in nature [
6], (d) runs in families [
7], (e) can be separated from other comorbid disorders on a number of relevant variables [
8‐
11] and, (f) correlates with biomarkers of aggression and impulsivity [
12]. Despite this, clinicians and researchers working with children and adolescents largely focus on DBDs in the context of anger, impulsivity, and aggression. Making matters more complex, a new disorder in DSM-5, codified as disruptive mood dysregulation disorder (DMDD; [
13]) also highlights anger and aggression, though DMDD is primarily conceptualized as a mood disorder. The primary difference between DMDD and IED is that the former represents a severe form of mood disorder in which anger is present most of time occurring before the age of ten while the latter describes individuals in whom aggressive outbursts are frequent but episodic and in whom anger is not present most of the time between outbursts. While not perfectly aligned with children and adolescents, studies in adults suggest that IED is comorbid with DMDD in less than 10% of cases [
14] indicting that the two may well be clinically separable.
Clinically, IED and the DBDs may be compared and contrasted in the following ways: (a) ADHD and IED share high levels of impulsive behavior but those with IED manifest serious aggression toward others, which is not characteristic of those with ADHD; in addition those with IED do not experience problems with sustaining attention as seen in ADHD; (b) CD and IED share history of aggressive behavior but this behavior tends to be anger-based/impulsive in IED but predatory/premeditated in CD; and (c) ODD and IED share history of temper tantrums but these are more frequent, and accompanied by more severe aggressive outbursts, in those with IED.
In this paper, we study the comorbidity of IED and DBD based on available empirical data from two large community surveys and from a relatively large clinical research data set. Disruptive Mood Dysregulation Disorder (DMDD) was not included in this study because of none of the data sets collected the data needed to make this diagnosis. We examined several aspects of comorbidity in IED, as well as the relative ages of onset of IED and DBDs. Since reliability is highest when considering concurrent diagnoses, we focused on those conditions present within the last 12 months for most analyses. Additionally, we also examined the quantitative nature of aggression as a function of IED and comorbid DBDs in the two community samples. In the Clinical Research sample we were able to examine the quantitative nature of DBD behavior scores as a function of IED and comorbid DBD. We hypothesized that while current IED would display an increased rate of comorbid DBD disorders, individuals with IED would display: (a) no more overall current comorbidity than those with DBD disorders, (b) ages of onset of IED precede that of each comorbid DBD disorder, (c) similarly elevated aggression scores in those with IED only, those with IED and each DBD, those with DBD only, and (d) elevated DBD scores only in those with DBD and with IED and DBD.
Methods
This study analyzed data from three sources. First, the National Comorbidity Survey-Adolescent Supplement (NCS-AS; [
15]); second, the National Comorbidity Survey -Replication of adults (NCS-R; [
16]) and, third, an adult clinical research sample engaged in research studies approved by the University of Chicago Institutional Review Board. The primary data source for this work was the NCS-AS with data from the NCS-R and the clinical research sample serving as a comparison between adolescents and adults. The two NCS data sets are publically available (NCS-AS:
http://dx.doi.org/10.3886/ICPSR28581 and NCS-R:
http://www.icpsr.umich.edu/icpsrweb/ICPSR/studies/20240).
Study samples
The NCS samples constitute surveys of mental disorders in the United States of America. The NCS-AS is made up of 10,148 adolescents of both sexes (mean ± SD age: 15.8 ± 1.5); the NCS-R involved 9282 adults of both sexes (mean ± SD age: 44.7 ± 17.5 years). Details regarding methods of data collection have been published [
15,
16].
Clinical research sample
The Clinical Research sample contained 1644 adults (mean ± SD: 33.3 ± 9.9 years) of both sexes who had completed at least one study in which research diagnostic/personality trait assessments were completed. Individuals were recruited from the community using public service announcements seeking participants for the various studies. Details regarding the clinical research sample have been published [
17].
Diagnostic assessments
The NCS surveys were designed to yield psychiatric diagnoses according to the DSM-IV [
18]. However, raw data in the NCS-AS/NCS-R data bases allowed diagnoses to be updated to DSM-5 [
19]. For the IED diagnosis, participants reported at least three “anger attacks” in any given year (Criteria A
2) and met the remainder of the DSM-5 criteria for IED. Although DSM-5 also allows frequent, but low intensity “anger attacks” (Criteria A
1), the NCS-AS/NCS-R surveys did not record data related to this type of “anger attack”. Psychiatric diagnoses in the Clinical Research Sample were made using DSM-5 criteria (Criteria A
1 and A
2) as previously described [
17]. Study participants in this sample with any psychiatric diagnosis (n = 1189), 58% (n = 690) reported a history of formal psychiatric evaluation and/or treatment in 58% (n = 690) of cases; an additional 14% (n = 166) of cases reported a history of behavioral disturbance during which the participant or others thought the participant should have sought mental health evaluation/treatment but did not. The NCS-AS study included data from interviews with the adolescents and from questionnaires about the adolescent by their parents in most (68%), but not all, cases; such informant data was not collected for NCS-R and Clinical Research samples.
Dimensional variables relevant to IED
Both NCS-AS/NCS-R surveys included six [
6] questions very similar to those from established assessments of aggression (e.g., “I have temper tantrums” compared with “I have trouble controlling my temper” from the Buss-Perry Aggression Questionnaire: BPAQ [
20] and impulsivity (e.g., “Giving into urges gets me into trouble” compared with “Do you often get into a jam because you do things without thinking?” from the impulsivity scale of the Eysenck Personality Questionnaire [
21] which enabled the creation of a variable for impulsive aggression. The scoring for the two surveys differed because the six NCS-AS items had four anchor points (0, 1, 2, or 3) and the six NCS-R items had two anchor points (0 or 3). The Clinical Research data contained aggression scores from the life history of aggression (LHA; [
22]) and the verbal and physical assault scores from the Buss-Perry Aggression Questionnaire (BPAQ; [
20]), assessments. Psychometric properties for LHA Aggression (e.g., α = 0.88) and for BPAQ Aggression (e.g., α = 0.85 for Physical, and α = 0.73 for Verbal, Assault) are good to excellent.
Dimensional variables relevant to DBD
The NCS-AS and NCS-R surveys did not include dimensional variables relevant to the severity of DBD disorders. While symptom counts for each DBD could be calculated, the structure of the interviews did not allow for an assessment of all DBD criteria in all subjects and, thus, could not be used. While this was also true for the Clinical Research group, data from the Wender-Utah Rating Scale (WURS [
23]), a Likert-scaled questionnaire assessing current and lifetime DBD (and other behaviors) were available in a sizable subset of the Clinical Research study participants (n = 713). The WURS contains twenty items that assess current severity of ADHD (separate scores for Hyperactivity-Impulsivity and Inattention), ODD, and CD, behaviors.
Statistical analysis
Statistical procedures included binary logistic regression for adjusted odds ratios, analysis of covariance (ANCOVA), and paired t-tests, as appropriate. All reported data was adjusted for age, sex, ethnicity, and education (level for parent for NCS-AS; level for subject for NCS-R) or Hollingshead Socio-Economic Status score (clinical research group). A two-tailed alpha of 0.05 was used to denote statistical significance for all analyses with Bonferroni-correction as appropriate. The first set of analyses involved examined the number of current disorders for each sample. This was followed by an examination of the rates (percentages) and risk (odds ratio) for overall comorbidity (e.g., comorbidity of a disorder with all other disorders). Next, we examined the rates and comorbidity risk for each DBD disorder as a function of IED taken separately as well as examining the comorbidity risk for all disorders in the same statistical model to determine the true comorbid nature of IED. The second set of analyses examined the age (and relative sequence) of onset for IED and each DBD disorder to determine the temporal nature of IED comorbidity. The third set of analyses examined mean aggression scores as a function of comorbidity. For example, subjects in each sample were divided into those with no life history of any disorder, those with a Non-IED/DBD disorder, those with a DBD disorder (e.g., ADHD, ODD, CD), those with IED, and those with both IED and DBD. This was performed to determine if aggression scores were higher in IED compared with those with DBD, and compared with both IED and a DBD (e.g., IED + ADHD). Composite Aggression scores were created for the Clinical Research group by taking the mean z scores for LHA and BDHI scores.
Discussion
Reanalysis of data from three different samples strongly suggests, with few exceptions, that: (a) comorbidity of current IED with current DBDs is similar to (or less than) current comorbidity of DBD disorders with other disorders; (b) taking all examined disorders simultaneously, current IED is significantly comorbid with each DBD with an odds ratio of about two; and, (c) mean aggression scores are highest among those with IED+DBD followed by those with IED alone, DBD alone, Psychiatric Controls, and Healthy Controls suggesting that the comorbidity of a DBD with the presence of IED is associated with an even greater severity of aggressive behavior than of IED alone. Notably, the reverse was not observed. When examining severity of DBD across the groups we observed that elevated DBD scores are characteristic of only those with DBD regardless of whether they also had IED. This suggests that DBD symptoms are unlikely to explain the aggressiveness of those with IED.
Examination of the temporal relationship of the disorders is important in understanding links between IED and DBD. The results were clear for ADHD, but less so for ODD and CD. In all three samples, the reported onset of ADHD was several years before that for IED, suggesting that the presence of ADHD increases the risk for developing IED at a later time. When examining both current and past diagnoses, the presence of IED persisted beyond the time of active ADHD in nearly half of adolescent cases (45.7%) and in the vast majority of adult cases (88.6%). Thus, despite the earlier onset of ADHD, the two disorders can be distinguished over time. Finally, while the risk of IED in those with ADHD is greater than in those without ADHD, only a quarter of adolescents (24.5%) with lifetime ADHD were comorbid for lifetime IED, indicating that lifetime comorbidity with IED does not account for most cases of ADHD. The reverse was also true with about an eighth of adolescents with lifetime IED (12.7%) having lifetime ADHD.
For ODD and CD, examination of the NCS-AS sample revealed that IED manifests itself before ODD in a plurality of cases (45.9%) and before CD in the majority of cases (69.3%). This was not true in the NCS-R and Clinical Research adult samples where the proportions of IED occurring first were lower, or about the same, as that occurring after ODD or CD. Given that the individuals in the NCS-AS sample were adolescents at time of study, and that reported history of psychopathology would be less affected by retrospective assessment in this sample (compared with the adult samples), one may give more weight to the results from the NCS-AS adolescent sample and suggest that IED manifests earlier than ODD and CD in more cases than not. Similar to ADHD, active IED was present when ODD (66.2%) or CD (78.0%) was not active, indicating that IED and ODD or CD can also be distinguished from IED over time. Finally, while the presence of IED may increase the risk of developing ODD or CD, less than a third with lifetime IED had lifetime ODD (29.2%) and less than a fifth had lifetime CD (18.3% %), indicating that lifetime comorbidity with ODD or CD does not account for most cases of IED.
Taken together, these data support the hypothesis that IED is a discrete disorder in adolescents in the same manner that DBDs are considered discrete disorders. That is, IED is not excessively comorbid with other current disorders to render it better explained by the presence of other another disorder or psychopathology, it occurs relatively early in life, and elevated aggression scores are characteristic of IED with or without the presence of a comorbid DBD disorder. Since aggressive behaviors in childhood and adolescence are associated with multiple undesirable outcomes, including juvenile delinquency, academic failure, and substance abuse, identifying IED, and making a proper diagnosis early in childhood, might provide a developmental opportunity to intervene and mitigate risk factors associated with aggression [
24].
This study has strengths and limitations. First among strengths, these results are based on a reanalysis of two large population-based community data sets and one relatively large clinical research data set. Second, diagnoses were updated to those of DSM-5, though only the A2 criteria for IED were applied (because questions relevant to the A1 criteria were not included in the survey instruments used at the time). That said the clinical research data set assessed IED by both A1 and A2 criteria and the results of these analyses rendered similar results as those with the community survey data set. Third, we were able to assess a variable for aggression in all samples and a variable for DBD severity for the clinical research sample, and found similar results.
This report differs from a previous report using the NCS-AS data set. First, the McLaughlin et al. [
3] report did not include all subjects who entered the NCS-AS study. This report included only participants 17 or younger and only participants with data from parent informants. Instead, we included 598 participants aged 18 and we included 3067 participants without data from paired informants. That said, our general results regarding prevalence of IED and the DBDs are similar suggesting that adding these data did not materially affect the reported findings.
This examination has limitations as well. First, the community sample data set was collected in the early 2000s and there may have been changes in the community-based epidemiology of IED and the other disorders examined. Unfortunately, we are not aware of another relevant community data targeted community survey to take place. Second, self-reported data is subject to retrospective bias and the presence or absence of disorders and the timing of onset of disorders could be affected by this factor. This is why we largely limited this analysis to examining current/past year disorders. Third, while two-thirds of the NCS-AS sample had parent informants, no informant interviews were conducted in the other two data sets. While desirable, informant interviews were not possible due to the expense this would have entailed. Fourth, our data regarding aggression severity in the community sample was derived from a group of personality items and not from full assessments of impulsive aggression as in the clinical research sample. That said, these items were drawn from established measures and results were consistent with that in the clinical research sample.
The present study adds to the growing body of literature on the comorbidities of IED. We emphasize that child and adolescent psychiatrists should think about IED in the deferential of DBD because early recognition of IED may help guide the treatment of aggressive behavior in such individuals. It is important to highlight that the presence of IED does not appear to alter the severity of DBD scores.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.