Helping Give Away Psychological Science/1234 Converting AIR Criteria to Wiki Preprint

Title Page
Title: Finding a Needed Diagnostic Home for Children with Impulsive Aggression

Ekaterina Stepanova, MD, PhD, Joshua A. Langfus, MA, Eric A. Youngstrom, PhD, Spencer C. Evans, PhD, Joel Stoddard, MD, MAS, Andrea S. Young, PhD, Kathryn Van Eck, PhD, Robert L. Findling, MD, MBA


 * Ekaterina Stepanova, MD, PhD (Q91540088)
 * Affiliation: Virginia Commonwealth University, Department of Psychiatry
 * ORCID: https://orcid.org/0000-0003-4033-2280
 * Email: ekaterina.stepanova@vcuhealth.org
 * Conflict of Interest Declaration: In the past 24 months, Dr. Stepanova receives or has received research support from Psychnostics, LLC and NIMH
 * Joshua A. Langfus, MA (Corresponding Author; Q103815757)
 * Affiliation: University of North Carolina at Chapel Hill, Department of Psychology & Neuroscience
 * ORCID: https://orcid.org/0000-0003-1096-966X
 * Email: langfus@unc.edu
 * Conflict of Interest Declaration: Mr. Langfus reports no real or potential conflict of interest
 * Eric A Youngstrom, PhD (Q27734333)
 * Affiliation: University of North Carolina at Chapel Hill, Department of Psychiatry & Neuroscience
 * ORCID: https://orcid.org/0000-0003-2251-6860
 * Email: eay@unc.edu
 * Conflict of Interest Declaration: In the past 24 months, Dr. Youngstrom receives or has received research support, acted as a consultant and/or has received honoraria from Pearson, Janssen, Joe Startup Technologies, Western Psychological Services, American Psychological Association, Guilford Press
 * Spencer C. Evans, PhD (Q56332127)
 * Affiliation: University of Miami, Department of Psychology
 * Email: sevans@miami.edu
 * Conflict of Interest Declaration: Dr. Evans reports no real or potential conflict of interest
 * Joel Stoddard, MD, MAS (Q64611623)
 * Affiliation: University of Colorado Anschutz Medical Campus, Aurora
 * Email: Joel.Stoddard@cuanschutz.edu
 * Conflict of Interest Declaration: Dr. Stoddard receives or has received grant or research support from the National Institute of Mental Health and Brain and Behavior Research Foundation. He has served as a DSMB committee member: Threat Interpretation Bias as Cognitive Marker and Treatment Target in Pediatric Anxiety (R61 Phase).
 * Andrea S. Young, PhD (Q38547062)
 * Affiliation: Johns Hopkins University, Department of Psychiatry and Behavioral Sciences
 * ORCID: https://orcid.org/0000-0002-8486-0643
 * Email: ayoung90@jhmi.edu
 * Conflict of Interest Declaration: In the past 24 months, Dr. Young receives or has received research funding from NIDA, the Brain and Behavior Research Foundation, Psychnostics, LLC, and Supernus Pharmaceuticals, has served as a consultant/grant reviewer for PCORI, NIH and Montana State University, serves on editorial boards for American Psychological Association journals, and on the advisory board for Helping Give Away Psychological Science (501c3).
 * Kathryn Van Eck, PhD (Q87955302)
 * Affiliation: Johns Hopkins University and Kennedy Krieger Institute
 * Email: vaneckk@kennedykrieger.org
 * Conflict of Interest Declaration: In the past 24 months, Dr. Van Eck receives or has received research support from Supernus Pharmaceuticals, Center for Disease Control, Health Resources and Services Administration, the National Institute of Child Health and Human Development.
 * Robert L. Findling, MD, MBA (Q90314248)
 * Affiliation: Virginia Commonwealth University, Department of Psychiatry
 * Email: Robert.Findling@vcuhealth.org
 * Conflict of Interest Declaration: In the past 24 months, Dr. Findling receives or has received research support, acted as a consultant and/or has received honoraria from Acadia, Adamas, Afecta, Akili, Alkermes, Allergan, American Academy of Child & Adolescent Psychiatry, American Psychiatric Press, Arbor, Axsome, Emelex, Gedeon Richter, Genentech, Idorsia, Intra-Cellular Therapies, Luminopia, Lundbeck, MedAvante-ProPhase, NIH, Neurim, Otsuka, PaxMedica, Pfizer, Q BioMed, Receptor Life Sciences, Roche, Sage, Signant Health, Sunovion, Supernus Pharmaceuticals, Syneos, Syneurx, Takeda, Teva, and Tris.

Abstract
Aggressive behavior is one of the most common reasons for referrals of youth to mental health treatment. While there are multiple publications describing different types of aggression in children, it is still challenging for clinicians to diagnose and treat aggressive youth, especially those with impulsively aggressive behaviors. The reason for this dilemma is that currently several psychiatric diagnoses include only some symptoms of aggression into the criteria. However, no diagnosis adequately captures youth with impulsive aggression (IA). Here we review the current diagnostic categories, including behavior and mood disorders, and show that they do not provide an adequate description of youth with IA. We also specifically focus on the construct of IA as a distinct entity from other diagnoses and propose an evidence-based set of diagnostic criteria that describes youth with IA to use for future evaluation in clinical practice.

Keywords: Developmental Psychology, Toddler, Medical Diagnosis, ICD criteria

Introduction
Aggressive behavior in young children is often considered to be a typical part of early development. Parents of toddlers are generally prepared to deal with the “terrible twos” expecting that aggressive behaviors will subside as the child grows older. In a vast majority of children, aggression begins to subside after their 3rd birthday. However, when aggression becomes impairing to the point of interfering with the child’s and family’s functioning, parents may seek help from a mental health provider. The job of the mental health provider during an evaluation becomes to accurately diagnose the child that presents with aggressive behaviors in order to recommend appropriate treatment. Unfortunately, no diagnosis in the DSM5 or ICD11 adequately captures aggression in children. The purpose of this review is to examine the limitations of the current nosology in describing children with aggression, specifically focusing on impulsive aggression (IA).

Aggression is one of the most common reasons for referrals to pediatric mental health clinics, emergency room visits and admissions to inpatient units. Historically, aggression has been divided into proactive (premeditated, instrumental goal-directed behavior that occurs with little or no provocation) and reactive (in response to a stimulus, such as a frustrating event or a perceived threat) types   with many individuals exhibiting elements of both. Reactive aggression is often described as “angry”, “hostile” or “hot,” while proactive aggression is commonly referred to as “cold” or “premeditated”. These two types of aggression have different biological and developmental correlates, as well as longitudinal course, and are likely distinct phenomena. Studies showed a correlation between proactive aggression and the development of delinquency and psychopathy later in life, while reactive aggression was associated with internalizing problems (including depression and anxiety symptoms), negative affect, ADHD, and peer problems. However, other studies report that reactive aggression may precede proactive aggression during the course of child’s development, suggesting a more complex interaction and co-occurrence between the two forms of aggression. This overlap is likely due to a shared neural basis, that is both types of aggression involve regions that mediate evaluating the subjective value of actions in any situation, social threat, and affective regulation.

For the purpose of this review, we take it that reactive aggression and IA describe similar, if not identical behaviors; however, we use the term IA because of recent work linking impulsivity and aggressive behavior in the context of ADHD, as we will elaborate further below. Here we primarily focus on children with impairing IA that’s inconsistent with their developmental level. In our experience, youth with debilitating symptoms of IA are often challenging to correctly diagnose and effectively treat. Families often describe aggressive behaviors as “outbursts”, “tantrums”, or “rages”. These children can have aggressive episodes several times a week, often daily, that include verbal and even substantial physical aggression. The severity and duration of these episodes vary greatly and do not seem to be predictable. These children may snap at a peer’s comment or destroy their room in a fit of rage. Their aggressive behavior is impulsive in nature and is triggered by events that are perceived as minor or insignificant by others. In the absence of triggers, however, these youths are generally calm. Nevertheless, the aggression can be so impairing that it leads to expulsion from daycare or school, causes difficulty functioning at home, negatively affects peer relationships, contributes substantially to caregiver burden, and even increases the rates of suicidal behaviors and ideation, all of which lead to seeking mental health treatment.

Although mental health providers frequently encounter and assess youth with impairing IA, there is no clear diagnostic framework to capture such behaviors. These children are therefore described as “diagnostically homeless”, despite symptoms of IA cutting across several extant diagnostic categories (DSM-5 and ICD-11). In particular, symptoms of impulsivity and aggression appear in the criteria for conduct disorder (CD), intermittent explosive disorder (IED), and disruptive mood dysregulation disorder (DMDD), and are often associated with attention-deficit/hyperactivity disorder (ADHD), oppositional defiant disorder (ODD) and other mood disorder diagnoses. Below, we suggest that these categories, as they are currently defined, do not sufficiently capture the phenomenology of IA. For that reason, our group evaluated the construct of IA and determined that it is distinct from attention, mood and rule-breaking problems. Here we propose to address the lack of an adequate diagnosis for children with IA by suggesting a set of diagnostic criteria to capture IA construct.

Conduct Disorder
Early research on childhood onset CD focused on identifying and treating youth with “severe aggressiveness, explosiveness, and disruptiveness unresponsive to various outpatient treatments”. Other studies have focused on a group of youngsters with early onset CD and ADHD that fit the same criteria. This description temporarily created a diagnostic home for those impulsively aggressive youth who frequently present for treatment. With the transition from DSM-III to DSM-III-R, however, more symptoms of proactive aggression and other behaviors (such as truancy, running away, using a weapon) were included, gradually shifting the diagnostic criteria away from a focus on impulsive aggressive symptoms. Over the transition from DSM-III to DSM-IV, the antisocial behavior presentations that were then captured by different subtypes of CD came to be codified in a single CD category defined by a heterogeneous array of problem behaviors. These criteria (in DSM-5 there are 15, of which only 3 are needed for diagnosis) are classified not only as aggression—which is our focus here—but also include acts of deceitfulness/theft and serious rule violations. As this modern conceptualization of CD crystalized in DSM-IV and beyond, so too did it become clear that CD and ODD (discussed below) were both clearly distinct and closely related, with CD being the more severe of the two. Perhaps due to this evolution, including the increasing prominence of rule-breaking behaviors in CD criteria, the diagnosis became more stigmatizing. The focus on rule-breaking behavior also confounded legal with mental health issues. These behaviors were considered to be difficult to treat, which led to most insurances not covering treatment of CD, leaving providers to search for another billable diagnosis for youngsters with IA. Nevertheless, many if not most youngsters with IA likely also meet current criteria for CD. Despite this, CD no longer provides an accurate description of youth with IA.

Understandably, research on CD followed the diagnostic criteria and lumped together youth with IA and rule-breaking behaviors. However, it is possible that childhood onset CD encompasses a different set of behaviors compared to the adolescent onset form of the disorder. Although CD can be diagnosed in children as young as 4-5 years of age, the age of onset of aggressive behavior problems generally ranges from 4 to 10 years with a sharp decline of aggression after age 10. Additionally, some studies show that symptoms of CD in younger children may not be stable over time, and most children will no longer meet the full criteria for CD after 2-5 years. Thus, these studies suggest that a childhood-limited course of CD could be a hallmark of a diagnosis separate from the persistent course of CD. Additionally, there is evidence that many individuals with CD have family history of antisocial personality disorder, and often have comorbid anxiety and ADHD. However, the family and comorbidity studies did not separate proactive aggression from IA and it is unclear if youth with IA and proactive aggression share the same characteristics. Given that prior studies of childhood-onset CD provided inconsistent results in examining the of age of onset, presence of impulsivity alongside proactive aggression, and family history of ADHD or mood disorders, it remains an open question whether the current criteria for CD could represent a diagnostic home for youth with IA or if a separate category is needed.

Oppositional Defiant Disorder
As noted above, CD and ODD are distinct but related. Rates of comorbidity between the two disorders are high, and ODD frequently co-occurs with various other mental disorders. Further, ODD and CD share developmental pathways such that many youths with CD previously had ODD as a developmental precursor. Accordingly, clinicians and researchers typically do not focus solely on CD or ODD without any consideration of the other. Yet our attention to ODD here is not merely obligatory. Whereas CD includes many symptoms of proactive aggression (see above), ODD is characterized by reactive disruptive behaviors, akin to reactive aggression or IA. Specifically, ODD is defined by 8 symptoms (of which 4 must be present for the diagnosis), organized around at least two symptom dimensions: an angry/irritable dimension (temper; touchy; angry) and a defiant/headstrong dimension (argues; defies; annoys; blames; spiteful/vindictive.

Although the diagnostic criteria for ODD do not explicitly include aggressive behaviors, ODD symptoms are very highly correlated with aggression, especially with reactive aggression. Indeed, many empirical models of youth psychopathology consider oppositional and aggressive behaviors as being closely related to one another, both falling along the overt (rather than covert) dimension of disruptive behavior problems. However, clinical decision-making is categorical rather than dimensional. Thus, from a person-centered perspective, latent class/profile analyses support differential prediction of aggressive behavior as an outcome of different dimensions of ODD (irritable vs defiant). At the same time, the distinction between ODD dimensions should not be overstated, and linkages to aggression require further investigation. In sum, ODD is a disorder with high comorbidity, great heterogeneity in presentation and outcome, characterized by emotional and behavioral dysregulation. For many youths with ODD, IA can be an important part of the clinical presentation. However, ODD, as currently defined in DSM-5 and ICD-11, does not identify IA as such; nor does it capture youths with IA in need of clinical care.

Attention-Deficit Hyperactivity Disorder
At this time, aggression is not conceptualized to be a core symptom of ADHD in the current nosology. However, impulsivity is a core feature of ADHD, therefore it may not be surprising that children with this disorder are at a higher risk of exhibiting IA. About 54% of youth with ADHD showed clinically significant levels of aggression in the Multimodal Treatment Study of Children with ADHD. In fact, aggression is considered to be an associated feature of ADHD. Specifically, reactive aggression (or IA) showed a much stronger correlation with ADHD, compared to proactive aggression. The severity and frequency of aggressive behaviors further increase when ADHD is comorbid with other behavior disorders, such as ODD and CD. There is evidence that IA and ADHD are distinct constructs, however, the relationship between the two is not entirely clear and may involve emotional impulsivity leading to anger outbursts.

Despite the paucity of studies evaluating the nosology of ADHD and IA, a lot of aggressive children with ADHD are impaired and require treatment. For example, 26% of youth with ADHD and IA continued to have impairing symptoms of IA despite adequate treatment of ADHD. To address the dearth of data in clinical interventions for aggressive children with ADHD, several trials were designed and targeted aggressive behaviors in addition to the primary diagnosis. In a large NIH-sponsored trial Treatment of Severe Childhood Aggression (TOSCA) participants were included if they had a primary diagnosis of ADHD with comorbid ODD or CD, as well as high baseline levels of aggression. Some studies utilized similar methodology,  while others targeted aggressive behaviors in youth with primary diagnosis of CD     with or without comorbid ADHD or a combination of disruptive behavior disorders (DBD), including CD, ODD or DBD, not otherwise specified. Mood disorders, such as Major Depressive Disorder (MDD) or Bipolar Disorder (BD), were excluded in most of these trials. While the methodology of these studies was drastically different, all of them attempted to select a population of highly aggressive and impaired children with disruptive behavior disorders that do not have significant mood problems. Unfortunately, none of these trials separated aggression into proactive or reactive subtypes and it is unclear if only youth with IA were selected for these studies. However, the trials not only highlight efficacy (or lack of efficacy) of different psychopharmacological agents for management of aggressive behaviors, but also bring attention to this unique population of aggressive children without comorbid mood symptoms that do not yet have a diagnostic home.

IED
Another possible home for youth with IA is IED. As the definition implies, IED manifests as a failure to control aggressive impulses. Changes made to the IED criteria in the DSM-5 allow for more frequent verbal and physical aggression (up to 3 times a week) than did earlier iterations of the DSM. However, most prior research on IED used the DSM-IV criteria (3 major aggressive episodes in the past 12 months), which likely captures a different phenotype from aggression that occurs on an almost daily basis. Unlike the group of children with early onset CD and ADHD, described above, IED is frequently comorbid with mood, anxiety and substance use disorders. Some studies report additional comorbidity of IED with ADHD and ODD. Family history of individuals with IED also differs from that of CD. First degree relatives generally have mood disorders, substance abuse and other impulse control disorders. In addition, available data suggest the age of onset of IED ranges from around adolescence to early adulthood rather than early childhood69, setting IED apart from ADHD and CD. To our knowledge, there have not been prospective studies examining the prevalence and persistence of IED in preschool or elementary school children. Although IED and IA appear to have a similar behavioral profile, the combination of other attributes substantially differ between IED and IA. Those include adolescent age of onset of IED, family history of mood disorders and comorbidity with mood and other impulse-control disorders, all of which make IED an unlikely nosological entity for kids with IA.

Mood Disorders
Aggression and irritability can also occur in the context of mood disorders. Severe outbursts were once hypothesized to be a hallmark of the pediatric BD, drastically increasing its rate of diagnosis in youth. Subsequently, it was shown that most irritable and aggressive children do not develop BD later in life, in contrast to those with more episodic presentations. To find an alternate “home” for youth with irritability, DSM-5 introduced a diagnosis of Disruptive Mood Dysregulation Disorder (DMDD), which specifically focuses on chronic irritability and frequent temper tantrums. However, in BD and DMDD (both of which DSM considers mood disorders), aggression tends to occur specifically in the context of changes in mood. In contrast, many children with IA appear angry during relatively brief episodes but are otherwise generally euthymic.

DMDD was revised from its research precursor, Severe Mood Dysregulation, largely by removing criteria for hyperactivity/arousal. The revision left DMDD to solely focus on a single, core construct, irritability, which is defined by two indicators: temper outburst and “irritable”/angry mood in between outbursts. Despite emerging from investigations on mood disorders, the revision brought DMDD close to chronic irritability as it is defined in the aggression literature. In current investigations, the degree to which these two indicators of irritability co-occur in clinical populations is being tested. Recent evidence suggesting they each may be individually present at a clinical significant level, i.e. they are separable. IA is promising in that it may be diagnosed when there is no evidence of chronic, between outburst negative mood. However, clinicians will need to take care in assessment, considering known issues by developmental stage and informant that obscure the detection of between outburst negative affect.

Aggression as a Construct
While symptoms of IA cut across several psychiatric diagnoses, none of the aforementioned diagnostic categories accurately capture phenomenology of IA. Aggression can, at times, be adaptive or maladaptive; it can present with a variety of psychiatric diagnoses or in the absence of psychopathology. There is not yet a consensus on whether maladaptive aggression should be a criterion of several diagnoses, an associated feature, a separate disorder, or all of the above. Connor et al began to answer that question by examining aggression in a group of psychiatrically referred children compared to healthy controls. They found that aggression scores were much higher in referred children with a variety of diagnoses, including disruptive behavior, anxiety and mood disorders, when compared to controls. In this study different types of aggression, including reactive and proactive, were highly correlated with each other, which supported the authors’ suggestion that there is a single distinct underlying aggressive construct that is transdiagnostic.

One recent effort to specifically focus on evaluating IA as a clinical construct occurred in a consensus conference with multiple experts, where the participants evaluated the research evidence available at that time. They concluded that there was indeed sufficient scientific evidence for the construct of IA, which can be measured and reliably recognized by experts. However, this construct is not specific to any particular diagnosis and can be present in youth with BD, depression, and ADHD. One possibility is that it constitutes a marker of severity. The authors of the report compared IA to non-specific symptoms such as pain or fever. The consensus conference helped advance the field’s understanding of IA and its boundaries in relation to other disorders.

Several years ago, our group began to work on validating the IA construct on a much larger scale using three data sets, as well as further clarifying the boundaries of IA as it relates to psychiatric diagnoses. Results from our work, which combined rational and empirical approaches, indicated that IA or Aggression with Impulsivity and Reactivity (AIR) is distinct from other mental health problems (such as mood problems, hyperactivity and rule-breaking behavior).

Forthcoming work from our group examined the conceptual boundaries of the AIR construct in the context of other mood and behavior symptoms. Drawing from the same three, heterogeneous samples as Young et al., provided access to a diverse group of participants from community and academic settings with different demographic characteristics (table 2). The latent profile analysis (LPA) identified groups of children with similar patterns of symptom elevation across five domains: AIR, depression, mania, rule-breaking and self-harm. Seven distinct symptom profiles replicated across the three samples. Of these, one had AIR as the predominantly elevated domain. Rates of ADHD diagnoses were higher in this group compared to sample averages, as were rates of ODD and CD. Children in this group were slightly younger than average, and they did not have elevated mood or anxiety concerns. Of the other profiles that emerged, several showed high levels of AIR co-occurring with mood and self-harm symptoms, and others showed mood symptoms without the presence of AIR.

Overall, this work supports three implications important for defining the AIR construct: first, a profile exists where AIR is the primary concern, establishing that AIR is not solely an associated feature of mood or disruptive behavior disorder; second, profiles exist where AIR does co-occur with mood and disruptive behavior diagnoses; third, AIR is not a necessary symptom for these diagnoses, since groups of children exist with these disorders and no elevation in AIR. Taken together, these results support a definition of AIR as both a distinct construct, and a transdiagnostic feature that co-occurs with other psychiatric conditions. These observations are therefore an important step toward identifying a diagnostic home for youngsters with IA, as well as an optimal setting for evaluating appropriate treatment.

To further the aim of creating a diagnostic home for youth with IA, we propose to develop data-driven research diagnostic criteria (RDC), and subsequently evaluate them in a clinical setting. At this time, our findings do not constitute a final set of clinical diagnostic criteria, but rather a starting point to move towards a potential new diagnosis. Establishing the RDC provides an opportunity for further evaluation of its applicability in clinical practice, as well as fine-tuning the criteria to accurately reflect the population of children with AIR. In addition, developing the RDC for youth with IA provides a framework for designing improved rating scales and other assessment tools for accurate identification of children with IA.

In order to establish preliminary RDC of AIR, we selected items included in the empirically-developed AIR construct, such as “assaults people physically”, “makes threats to others”, “fights frequently with others”, “acts impulsively or rash”. Unfortunately, many of the aggression items on the assessments are not specific to IA and do not provide context. We, therefore, supplemented the available AIR items from the assessments with other criteria that in our opinion describes IA. For that purpose, we added a more thorough description of verbal aggression (“screams, yells” and “insults adults”) and physical aggression (“kicks, hits, bites”, “scratches”). We also added the description of aggressive behaviors being impulsive, rather than planned. These additional criteria, including frequency and duration of the behaviors, as well as the number of criteria used to make the diagnosis of AIR were added on the basis of extensive clinical experience and consensus among several experts. The provisional age of onset was informed by early research on CD.

Proposed: AIR Criteria

 * 1) Recurrent failure to control aggressive impulses resulting in frequent outbursts (verbal and/or physical) often in response to minimal triggers (such as being given directions, non-preferred tasks or comments that are perceived as critical). The outbursts are grossly out of proportion in intensity or duration to the situation. The outbursts include at least 3 symptoms out of 7 from the following categories, and not limited to interaction with one person, such as parent or guardian
 * 2) Verbal aggression
 * 3) Screams, yells; extraordinarily loud
 * 4) Insults adults
 * 5) Threatens others
 * 6) Physical aggression
 * 7) Hits, kicks, bites, pushes other people or animals
 * 8) Hits, bites, scratches him/herself
 * 9) Destruction of property
 * 10) Punches walls, kicks or hits furniture resulting in property damage
 * 11) Slams doors, throws small items or furniture
 * 12) Aggressive outbursts are inconsistent with developmental level
 * 13) Most aggressive behaviors occur at least 5 days out of a week
 * 14) Duration – at least 6 months
 * 15) Behavior occurs in at least two settings (ex. home and school) and is not restricted to the individual’s relationship with his/her parents or guardians
 * 16) Age of onset – prior to age 10
 * 17) Behaviors are not better explained by an episode of a mood disorder (depression or mania), anxiety disorder, psychotic disorder or ASD
 * 18) Symptoms are not attributable to the physiological effects of a substance or another medical/neurological condition
 * 19) The majority of recurrent aggressive outbursts are not premeditated
 * 20) The behavior causes clinically significant impairment in social, academic, or occupational functioning

The preliminary RDC of AIR listed above are a constellation of data-driven AIR-specific items, earlier research on childhood aggression (CD) and clinical experience of several experts (Fig. 1). While these RDC have face validity, we wanted to ascertain the clinical utility of AIR criteria. For this purpose, we applied the criteria to a small sample of patients presenting with a chief complaint of “aggression” at a community mental health clinic. Out of 15 patients with aggression, 7 met full criteria for AIR, while others were excluded due to IA occurring only in the context of a mood episode. We recognize the limitations of using such a small sample, however, this is the first step towards applying the criteria in a clinical setting.

It is apparent that these RDC of AIR resemble criteria for IED, which is a well-established diagnosis in the DSM-5 (Table 1). Naturally, we wondered if AIR is an early onset of IED in children. Unfortunately, there are no data available to date to answer that question. The symptom frequency criterion for IED changed significantly from the DSMIII to the DSM-5, with DSM-5 resembling AIR the most. However, to our knowledge, no prospective clinical trials evaluated the onset of the IED criteria in pediatric population and long-term outcomes in adulthood. Whether children with AIR grow up to be adults with IED is an empirical question that will need to be addressed in future studies.

In this proposal, IA in AIR is distinguished from temper outbursts in DMDD by requiring more frequent outbursts and defining the behavior as aggressive. Patient characteristics described in AIR such as explosiveness, sensitivity to provocation, having impairment in two domains (e.g. with family, school, or peers), and having affective dysregulation are similar to those in DMDD. Thus, the focus on aggression is important here as a potential for discriminant validity aside from lack of predominantly negative between-outburst mood. Aggression has most prominently been defined as a behavior intended to cause another person harm. Temper outbursts are intense, transient, negative affective expressions that often involve emotional, unplanned aggression and other affective displays. They have developmentally specific presentations, time course, and indicators for pathology. In the general population, trait aggression diverges from trait irritability and in pediatric clinical populations. More specifically, aggression is not necessarily observed during outbursts in clinical populations, where for example, property destruction, self harm, and negative self-talk are also reported by caregivers. In sum, AIR differs from DMDD in highlighting the importance of aggression during extreme affective reactions to provocation. This distinction has promise in that aggression is readily observable, especially when directed towards peers, and has predictive value for psychopathology.

Conclusion
Future research should examine the RDC of AIR in prospective field trials in a variety of clinical settings. These trials will provide additional data on the number of specific criteria needed to identify youth with AIR, as well as duration of illness, frequency of outbursts and age of onset, as these were selected via expert opinion and informed by criteria of other disruptive behavior problems rather than empirically defined. It is important to test these criteria in the general population of young children as well, since tantrums can occur as a part of normal early development. As comorbidity is usually the rule, rather than the exception, we recommend that future studies explore increases in impairment resulting from additive effects between AIR and other diagnoses. It is important to understand the longitudinal course and prognosis of AIR, including whether AIR in childhood leads to the development of IED later in adolescence or adulthood. Perhaps the most crucial question that needs to be answered is how to best help children with AIR and their families. TABLE 1''. Comparison between proposed AIR criteria to existing DSM-5 diagnoses of IED, CD, and DMDD''

Note: IED = Intermittent Explosive Disorder, CD = Conduct Disorder, DMDD = Disruptive Mood Dysregulation Disorder, AIR = Aggression with Impulsivity and Reactivity TABLE 2. Demographic characteristics of Stanley, ABACAB and LAMS samples

FIGURE 1. Overlap between symptoms of AIR and other diagnoses in the DSM 5

''Note. Each box in the middle represents one of the diagnostic criteria of AIR. Boxes and arrows on the left show overlap of symptoms between related diagnostic constructs and AIR. Dotted line represents the difference between specific behaviors in the diagnosis of CD and AIR.''

IED = Intermittent Explosive Disorder, CD = Conduct Disorder, DMDD = Disruptive Mood Dysregulation Disorder, AIR = Aggression with Impulsivity and Reactivity