Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

Consensus Report on Impulsive Aggression as aSymptom Across Diagnostic Categories in ChildPsychiatry: Implications for Medication Studies

PETER S. JENSEN, M.D., ERIC A. YOUNGSTROM, PH.D., HANS STEINER, M.D.,

ROBERT L. FINDLING, M.D., ROGER E. MEYER, M.D., RICHARD P. MALONE, M.D.,

GABRIELLE A. CARLSON, M.D., EMIL F. COCCARO, M.D., MICHAEL G. AMAN, PH.D.,

JAMES BLAIR, M.D., DONALD DOUGHERTY, PH.D., CRAIG FERRIS, PH.D.,

LAURIE FLYNN, B.A., EVELYN GREEN, B.A., KIMBERLY HOAGWOOD, PH.D.,

JANICE HUTCHINSON, M.D., TOM LAUGHREN, M.D., LESLIE D. LEVE, PH.D.,

DOUGLAS K. NOVINS, M.D., AND BENEDETTO VITIELLO, M.D.

ABSTRACT

Objective: To determine whether impulsive aggression (IA) is a meaningful clinical construct and to ascertain whether it is

sufficiently similar across diagnostic categories, such that parallel studies across disorders might constitute appropriate

evidence for pursuing indications. If so, how should IA be assessed, pharmacological studies designed, and ethical issues

addressed? Method: Experts from key stakeholder communities, including academic clinicians, researchers, practicing

clinicians, U.S. Food and Drug Administration, National Institute of Mental Health, industry sponsors, and patient and

family advocates, met for a 2-day consensus conference on November 4 and 5, 2004. After evaluating summary

presentations on current research evidence, participants were assigned to three workgroups, examined core issues, and

generated consensus guidelines in their areas. Workgroup recommendations were discussed by the whole group to reach

consensus, and then further iterated and condensed into this report postconference by the authors. Results: Conference

participants agreed that IA is a substantial public health and clinical concern, constitutes a key therapeutic target across

multiple disorders, and can be measured with sufficient precision that pharmacological studies are warranted. Additional

Accepted August 8, 2006.Drs. Jensen and Hoagwood andMs. Flynn are with Columbia University and

the New York State Psychiatric Institute, New York; Dr. Youngstrom is with theUniversity of North Carolina at Chapel Hill; Dr. Findling is with Case WesternReserve University, Cleveland; Dr. Steiner is with the Stanford University Schoolof Medicine, Palo Alto, CA; Dr. Meyer is CEO, Best Practice ProjectManagement, Inc., Bethesda, MD; Dr. Malone is with Drexel UniversityCollege of Medicine, Philadelphia; Dr. Carlson is with State University of NewYork, Stony Brook; Dr. Coccaro is with the University of Chicago School ofMedicine, Chicago; Dr. Aman is with the Nisonger Center, Ohio StateUniversity, Columbus; Drs. Vitiello and Blair are with the National Institute ofMental Health, NIH, Bethesda, MD; Dr. Dougherty is with Wake Forest HealthSciences, Wake Forest, NC; Dr. Ferris is with the University of MassachusettsMedical School, Worcester; Ms. Green is with the Attention Deficit DisordersAssociation, Chicago; Dr. Hutchinson is with Howard University Hospital,Washington, DC;Dr. Laughren is with the U.S. Food andDrug Administration,Rockville, MD; Dr. Leve is with the Oregon Social Learning Center, Eugene; andDr. Novins is with the University of Colorado School of Medicine, Denver.

The conference on which this article is based was funded by unrestricted grantsfrom Annie E. Casey Foundation, Bristol-Myers Squibb, Forest ResearchInstitute, Jazz Pharmaceuticals, Otsuka Pharmaceuticals, Janssen, Eli Lilly andCompany, Pfizer Laboratories, and Sanofi Synthelabo.

Dr. Vitiello_s and Dr. Laughren_s contributions to this article were made intheir private capacity. No official support or endorsement by the NationalInstitute of Mental Health or the U.S. Food and Drug Administration isintended nor should be inferred.

Additional consensus conference participants are gratefully acknowledged inalphabetical order: Jorge L. Armenteros, M.D., Rahn Bailey, M.D., Joseph Blader,Ph.D.,Daniel F.Connor,M.D., ReuvenFerziger,M.D.,VivianKafantaris,M.D.,Russell Katz, M.D., Stuart Kushner, M.D., Amy O_Donnell, M.D., Patrick T.McCarthy, Ph.D., Elizabeth Pappadapulos, Ph.D., Phillip Perera, M.D., KennethM. Rogers, M.D., Samuel Saks, M.D., Eugene Schneider, M.D., and CharlesWolleben, Ph.D. The authors acknowledge the advice of Andrew Leon, Ph.D.,Cornell University, for consultation about design and statistical issues, MelissaDelBello, M.D., for statistical analyses, and Diana Benzaia, M.A., for assistance inconference synthesis, editing, and manuscript preparation.

Article Plus (online only) materials for this article appear on the Journal_sWeb site: www.jaacap.com.

Correspondence to Dr. Peter S. Jensen, Center for the Advancement ofChildren_s Mental Health, Columbia University/NYSPI 1051 Riverside Drive,Unit #78, New York, NY 10032; e-mail: pj131@columbia.edu.

0890-8567/07/4603-0000�2007 by the American Academy of Childand Adolescent Psychiatry.

DOI: 10.1097/CHI.0b013e31802f1454

1J . AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 46:3, MARCH 2007

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

areas of consensus concerned types of measures, optimal study designs, and ethical imperatives. Conclusion: Derived

from scientific evidence and clinical experience, these consensus-driven recommendations can guide the design of future

studies. J. Am. Acad. Child Adolesc. Psychiatry, 2007;46(3):0Y0. Key Words: impulsivity, aggression, clinical trials,

assessment ethics.

In child and adolescent psychiatry the decision toinitiate, change, or increase medication dose is oftentriggered by severe aggression in the child or adolescentwho cannot be managed by caregivers. Patients withattention-deficit/hyperactivity disorder (ADHD)/dis-ruptive behavior disorders, bipolar disorder, childhoodpsychosis, autism and other specific developmentaldisabilities, and internalizing disorders such as post-traumatic stress disorder (PTSD) and depression maypresent with aggressive behavior that is difficult forpatients to contain and for caregivers to manage.Although there is an emerging literature on rando-mized, placebo-controlled trials in the treatment of anumber of disorders in child and adolescent psychiatryand U.S. Food and Drug Administration (FDA)approval for some medications in the treatment ofsome of these disorders, the construct of impulsiveaggression (IA) has not yet been fully agreed on orwidely recognized as an appropriate therapeutic target.Nonetheless, evidence indicates that aggression oftenco-occurs in specific disorders, but may not besufficiently ameliorated by standard medications usedto treat these disorders (e.g., ADHD; see Aman et al.,2004; MTA Cooperative Group, 1999). Because therehave been few rigorous clinical trials on the treatment ofaggression in children with well-defined DSM disorders(other than conduct disorder), clinicians are oftenobliged to use Bbest guesses[ for appropriate treatmentstrategies when IA complicates the management ofADHD, autism, bipolar disorder, PTSD, and otherpsychiatric disorders. Although treatment guidelineshave been recently published concerning pharmacolo-gical management of aggression in the context ofvarious DSM-defined conditions (Pappadopulos et al.,2003), it is notable that almost all of the recentcontrolled trial literature on which these guidelineswere based was drawn from studies of children/youthswho were required to meet criteria for conduct disorder(Schur et al., 2003). Pharmacological studies of autismhave been the singular exception to this rule, andinvestigators have increasingly and successfullyaddressed IA and irritability as principal therapeutic

targets in autistic patients (Research Units on PediatricPsychopharmacology Autism Network, 2002).

A BRIEF REVIEW OF IA

Within appropriate contexts, aggression servesimportant survival and adaptation purposes. Allhuman beings may express various forms of aggressionon a frequent basis (from mild irritation to anger toverbal or physical aggression), but it can be maladaptiveand constitutes one of the most common referralproblems to child psychiatric clinics (Connor, 2002).Normal and maladaptive aggression overlap with, butare not identical to, delinquency (Connor, 2002,Steiner and Cauffman, 1998). Great care must betaken not to confound adaptive and maladaptivebehavior for theoretical, empirical, and ethical reasons(AACAP, 1997).Maladaptive aggression can be defined as aggression

that occurs outside an expectable social context. Theintensity, frequency, duration, or severity of theaggressive response is disproportionate to its causesand may occur in the absence of expectable antecedentsocial cues. The aggressive behavior is not terminated inan appropriate time frame and/or in response tofeedback (Connor, 2002). Youngsters with maladaptiveaggression have more school adjustment problems,greater deficits in cognition, and experience more peerrejection and victimization. They have difficulties inambiguous interpersonal situations, such as readingemotion in people_s facial expressions. In particular,they are far more likely to read a neutral expression asnegative (disgust, fear, or anger), in contrast to normalyouths who are more likely to read neutral faces aspositive (surprised, happy, or kind; Best et al., 2002;McClure et al., 2003). In summary, maladaptiveaggression can be reasonably well characterized as asymptom and can be differentiated from the willful orplanned violation of societal norms, a form ofaggression that serves an adaptive purpose (at leastfrom the perspective of the perpetrator).Basic neuroscience, epidemiological, and clinical

studies provide considerable support for the existence

JENSEN ET AL.

2 J . AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 46:3, MARCH 2007

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

of two subtypes of aggression. IA is unplanned andovert. The aggressor perceives the outcome of theaggressive act to be negative with negative accompany-ing emotions, such as frustration, regret, guilt, and fear(Donovan et al., 2003). Although planned aggressionmay result in the outward expression of a keenly feltnegative emotion, such as disgust or contempt, unlikeIA it is often covert, with the perpetrator anticipating apositive outcome, such as heightened interest, satisfac-tion, or happiness (Steiner et al., 2003). Factor andother analyses support these two constructs as over-lapping (Malone et al., 1994) but relatively indepen-dent in school (Poulin and Boivin, 2000; Steiner et al.,2005a), clinical (Connor et al., 2004), and delinquent(Steiner et al., 2005b) populations, with IA morecharacteristic of clinical samples and covert or plannedaggression more characteristic of delinquent or criminalpopulations.

Epidemiological evidence also supports this dichot-omy. Developmental trajectories show antecedentspredicting problems with IA from late preschool ageforward, starting at about 4.5 years (Dodge, 1991;Dodge et al., 1997; Vitaro et al., 2002), including poorpeer relationships and inadequate problem-solvingpatterns. Studies suggest that 21% of children with IAhave a history of physical abuse (Dodge, 1991). IA isstable over time across the school years and becomesincreasingly associated with diverse and significantpsychopathology. In contrast, planned aggression startsat about 6.5 years and is associated with aggressive rolemodels in the family rather than physical abuse (seeDodge, 1991; Dodge et al., 1997; Porter, 1998).Children with planned aggression attribute positivevalence to aggression (Dodge et al., 1997).

Animal studies show that impulsive and proactiveforms of aggression are associated with differentpatterns of brain activation (Blair, 2004). Rodentmodels of aggression suggest that there may be differentneurobiological mechanisms associated with differentmodels of aggression, even though some medicationsmay be equally effective in damping down aggression,whatever the mechanism (Ferris and DeVries, 1997).Although there is mounting evidence that IA is aclinically distinct and important construct, associatedwith a wide range of psychopathology (Aman et al.,2004; Research Units on Pediatric Psychopharmacol-ogy Autism Network, 2002), to date it has not beenclear that aggression is the same symptom across

disorders in child psychiatry. Within specific disordersit has not been clear that the symptom of IA is discrete/separable in terms of treatment response from othersymptoms of the disorder, nor has it been clear whetherIA constitutes an appropriate and specific treatmenttarget.In spite of a lack of data from controlled clinical trials

of IA as a distinct treatment indication, the preponder-ance of children and adolescents hospitalized in a largestate inpatient system receive pharmacotherapy for thetreatment of aggressive symptoms, with 40% receivingtwo or more medications, most commonly an atypicalantipsychotic (Pappadopulos et al., 2002). Similarly,recent data from a representative national samplerevealed that 18% of all child/adolescent visits topsychiatrists in 2002 included the use of an atypicalagent, with the largest proportion of these visits forpatients with disruptive behavior disorders and aggres-sion (Olfson et al., 2006). Atypical agents appear to bemost common, however, other agents used in thetreatment of aggression in children and adolescentsinclude anticonvulsants, stimulants, selective serotoninreuptake inhibitors, anxiolytics, !-agonists, "-blockers,and sedatives (Connor, 2002; Steiner et al., 2003).Going forward, carefully executed clinical trials testingthe role of these agents for IA management are criticallyneeded to enhance the quality of clinical practice, tospare children from ineffective medications withaccompanying side effects, and, ultimately, to improveoverall outcomes.Studies of aggression in children raise important

ethical questions. Demographic and cross-culturalissues must be responsibly considered, so that studydesign, data, and conclusions cannot be used to impugnany group. Likewise, the role of well-establishedbehavioral or milieu interventions that have beenpreviously shown to be efficacious in aggressive children(Kazdin et al., 1989, 1992; Kellam et al., 1998;Lochman and Curry, 1986; Lochman and Lampron,1988; Malone et al., 1997) must be considered,potentially as a precondition for or concomitant ofpharmacological treatment.

METHOD

To define the field_s state of readiness to initiate programs ofresearch on IA treatment in children and adolescents with majorpsychiatric disorders, 40 experts in the field met at a 2-dayconsensus conference November 4 to 5, 2004, at which careful

IMPULSIVE AGGRESSION IN CHILD PSYCHIATRY

3J . AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 46:3, MARCH 2007

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

reviews of the literature and reanalyses of extant data sets werepresented on each of the topics listed inT1 Table 1.

Participants

Conference attendees included leading clinical investigators inchild and adolescent psychiatry, clinical scientists from majorpharmaceutical companies with antipsychotic and anxiolytic drugproducts (or products under development), and leaders fromgovernment agencies (e.g., FDA, National Institute of MentalHealth). Invitees also included parents and representatives fromgroups advocating on behalf of children and families, including theNational Alliance for the Mentally Ill, Children and Adults WithAttention-Deficit/Hyperactivity Disorder, and the Federation ofFamilies for Children_s Mental Health.

Meeting Format

During the first part of the conference, presentations laid out thestate of scientific and clinical knowledge on the construct of IA, aswell as the methodological and scientific challenges that need to beaddressed. After the presentations, participants were divided intothree workgroups and asked to address questions in each of threetopic areas: assessment of IA, design of pharmacological studies, andethical issues in pharmacological studies.

Assessment of IA.

• Is IA a sufficiently similar construct across diagnostic categoriesthat parallel studies across disorders or studies with broaddiagnostic inclusion criteria may constitute appropriate evidenceto pursue an indication?

• How should IA be assessed? What, if any, current measures maybe adequate for industry-sponsored studies of IA?

• Does intermittent explosive disorder (IED) Bwork[ as adiagnostic category in youths, or would youths with significant

clinical aggression be under- or overidentified? Although IED as adiagnostic category has rarely been used in studies of children, itsusefulness in adult studies (Barratt et al., 1999; Best et al., 2002;Blair, 2004; Coccaro and Kavoussi, 1997; Coccaro et al., 1991,2004; Kessler et al., 2006) may suggest the need to revisit thevalue of the greater use of this category in children and youths.

• What, if any, new assessment tools are needed? What studiesshould be done to validate these tools, and how should suchstudies be designed? How should such tools be studied acrossdifferent age, ethnic, or gender groups?

• What other constructs should be assessed in IA and why?

Although conduct disorder has historically been associated withaggression, this diagnosis has been controversial and has been shownto be influenced by other factors, including hidden causalattributions held by diagnosticians when they become aware ofadditional environmental factors (Hsieh and Kirk, 2003; Wakefieldet al., 2002). For these reasons, the conference objectivesdeliberately focused on disorders other than conduct disorder perse, although participants generally acknowledged and understoodthat many, perhaps most, children with IA may also meet criteria forconduct disorder. Somewhat less controversy has been attached tothe design of treatment studies for other childhood mentaldisorders, such as depression, bipolar disorder, autism, and evenADHD, where the FDA has encouraged pharmacological studiesand has approved indications for medications to treat theseconditions in children and/or adults and where studies havesuggested that the disorder may not simply be a function ofenvironmental stressors (Rey et al., 2000).

Design of Pharmacological Studies.

• Should IA be studied across diagnoses, powered to test for theeffect of diagnosis or within diagnosis (e.g., aggression within adisorder)? Which disorders would be most relevant andinformative regarding the treatment of IA?

• Given multiple comorbidities in children and youths, how doesone choose the primary disorder to be addressed? Should anydisorders be excluded? Should developmental disorders (e.g., IQbelow 70 or 80) be studied separately?

• Should aggression be targeted only after treatment of the primarydisorder has proven unsuccessful in sufficiently mitigating theimpairment due to aggression? How vigorous should treatment ofthe primary condition have been to assume that it has failed?Should aggression treatment be an Badd on[ to whatever othermedications or psychotherapeutic treatments the child is alreadyreceiving?

• Should IA be defined to have a certain duration (e.g., >6months)? Should it occur in more than one setting (home onlyversus home and school versus school only)?

• Study population issues were addressed. How should age andgender be handled? Do any other demographic factors need to beconsidered separately? Should children in hospitals, residentialsettings, or foster care be targeted or disallowed?

• What should be the length of clinical trials, and how might thatdiffer for different classes of agents? Should studies contain bothacute and discontinuation phases?

• Are there particular agents or classes of agents that should be prioritizedby Bwritten request[ from the FDA for future studies of IA?

• With regard to safety issues, how should behavioral (includingcognitive) toxicity be addressed? Should a priori definitions forimportant behaviors such as activation and suicidality be used?

TABLE 1Presentation Topics

Overview of impulsive aggression (IA) in children and adolescents:what is it, who is being treated, for what conditions, with whichmedications, at what doses?

The psychobiology and neuropharmacology of aggression inanimals: implications for children and adolescents with mentaldisorders

The clinical presentation(s) and treatment response of IA in specificdisordersADHDAutism/pervasive developmental disorder spectrumBipolar disorder

Clinical presentation(s) and measurement of IAPsychometric characteristics of the construct of IAA convergent analysis of IA across psychiatric disorders? Does thepain/fever analogy apply?

Critical perspectives on the measurement and treatment of IAPsychotherapeutic treatments and the role of medicationParental perspectivesChild policy perspectivesEthical perspectives

JENSEN ET AL.

4 J . AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 46:3, MARCH 2007

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

For medications known to produce weight gain, shouldadditional treatments, dietary management strategies, or otherapproaches be used to mitigate those problems?

Ethical Issues in Pharmacological Studies.

• Are any special ethical protections needed to ensure that childrenare not inappropriately placed in IA medication trials? What arethese special considerations in terms of possible inclusion/exclusion criteria?

• Are special precautions or other preparatory steps needed beforechildren are placed in such trials (e.g., psychotherapeutic orenvironmental interventions)?

• In terms of the riskYbenefit balance, are there any populations atspecial risk of inappropriate inclusion in studies of IA, and, if so,what are they (e.g., youths in juvenile justice settings or underlegal proceedings, ethnic minorities, or children in foster care)?

• Are placebo trials appropriate in youths with IA who have apreexisting condition such as ADHD/disruptive behavior dis-orders, autism, or bipolar disorder, or should trials only be Badd-on[ studies?

• Should any special considerations be taken to assess IA differentlyfor various cultural or ethnic groups?

Because not all of the above questions could be addressed in the timeavailable, each group determined its highest priority questions,focusing on those with the potential greatest importance for thefield. Following the workgroups_ deliberations, recommendationsfrom each group were presented to the overall conference, and theserecommendations were then further iterated to reach expertagreement about what is known, where there is consensus, andhow the field should proceed with future pharmacological studies.These draft recommendations were then further refined after themeeting by all attendees, and the final recommendations with theiraccompanying rationale were developed, as detailed below.

RESULTS

Consensus Findings and Recommendations

Recommendations were reached in most but not allareas. Recommendations are ordered below in terms oftheir overall priority as established by conferenceparticipants rather than in the order of the originalquestions.

IA, as has been measured and studied to date, is in fact asufficiently similar construct across diagnostic categories,such that parallel studies across disorders or broaddiagnostic criteria can and should be conducted. Suchstudies, if appropriately designed, are interpretable andappropriate to pursue medication indications.

For this conference, a review of measurementinstruments was conducted by researchers at CaseWestern Reserve University and Columbia Universityof recently published studies of aggression. After blindratings of items by three expert raters (P.J., R.F., E.Y.),analyses indicated that the IA construct shows up on the

Parent General Behavior Inventory, the Young ManiaRating Scale, the Aberrant Behavior Checklist, theNisonger Child Behavior Rating Form, and the ChildBehavior Checklist, with interrater agreement $ valuesof .73 to 1.00 among raters. Of note, IA is actually asmall subset of the items available on each of these scales( F1Fig. 1). IA items are largely found on the subscales forexternalizing behavior, aggression, or delinquent behav-ior (across these scales). Analyses indicated that an IAfactor with robust psychometric characteristics, goodinternal consistency, and high correlations among raterscan be derived from these diverse scales. (Supportingdata tables prepared by coauthor E.Y. are available onthe Journal_s Web site www.jaacap.com via the ArticlePlus feature.)Compared to normals, receiver operating character-

istic analyses completed on several separate studies anddata sets available to the authors indicated that IA issignificantly elevated in bipolar disorder, unipolardepression, and ADHD, but the symptom is notspecific to any of these conditions and is less specificthan the core symptoms of each disorder. Latent classanalyses indicated that 50% to 90% of cases with any ofthe above Axis I diagnoses (allowing comorbidity withother psychiatric diagnoses) are accompanied bymedium or high levels of impulsive aggressive behaviorsversus less than 10% of youths without an Axis Idiagnosis ( F2Fig. 2). (Also see supporting data tablesprepared by coauthor E.Y. via the Article Plus feature.)Thus, IA is a construct that reflects a single latent

variable that can be readily extracted from existing

Fig. 1 Percentage of impulsive aggression items of total number ofexternalizing items by scale. YMRS = Young Mania Rating Scale; P-GBI =General Behavior Inventory (Parent Version); CBCL = Child BehaviorChecklist; ABC = Aberrant Behavior Checklist; NCBRF = Nisonger ChildBehavior Rating Form; N = 480.

IMPULSIVE AGGRESSION IN CHILD PSYCHIATRY

5J . AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 46:3, MARCH 2007

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

measures in empirically validated scales. It can bereliably recognized by expert raters and measured withhigh internal consistency. IA scores converge into asingle factor across measures and raters. Although it is asensitive marker of severity of psychopathology, withsubstantial elevations in clinical versus nonclinicalgroups, it is not a marker of a specific diagnosis, andit is less specific than the core symptoms of the DSM-defined conditions in which it is found. In effect, IAappears to be analogous to fever or pain because it canbe reliably measured, is manifest across a variety ofdisorders, and is informative about the severity but notthe type of illness.Although the behavioral characteristics of IA can be

captured by similar measurements across dissimilar patientgroups, the likelihood of detecting a ‘‘signal’’ for anefficacious treatment is best achieved when clinical trialsfocus on the treatment of the symptom in patients withspecific DSM-defined disorders. Therefore, given thecurrent state of knowledge, for the purposes of pursuingpharmacological indications, IA should be studiedprincipally within well-defined patient groups with well-established psychiatric disorders, such as ADHD, PTSD,autism, and/or bipolar disorder.To the extent that other clinical conditions (e.g.,

pervasive developmental disorder [PDD], bipolardisorder not otherwise specified, personality disorders)can be well defined and operationalized, they also willwarrant study, given the likelihood that IA characterizesthese children as well. Each of these conditions posesunique challenges in study design. For example,

although IQ restrictions for these studies are generallynot needed, in studies of ADHD, it would make senseto undertake separate studies of youngsters withADHD characterized by normal IQ and those withlow IQ. Both groups would be worthy of study, even ifone limits the heterogeneity of the sample for thepurposes of a particular study.Participants concluded that the IA construct can

most fruitfully be studied within disorders such asADHD, autism, PTSD, and bipolar disorder (i.e.,those that are affectively overactivated and/or under-regulated). Such studies should allow the determinationof the extent to which IA accounts for overall clinicalimpairment, as well as the extent to which treatmentbenefits the primary disorder versus IA outcomes,illustrated at the conference and here below with threeexamples: ADHD, autism, and bipolar disorder. Therationale for studying IA within specific disorders isbased on the fact that for psychiatric disorders, the FDAusually approves medication indications first withinspecific disorders rather for a symptom that cuts acrossmultiple disorders. There are situations in which theFDA approves medications for symptoms not specificto any single disorder (e.g., fever, pain), even withinpsychiatry (e.g., agitation within dementia, suicidalitywithin schizophrenia), but in most instances, thepreferred approach is to first study a given medication_seffects on a symptom in a single disorder. If it can beshown that the same agent is repeatedly effective on thetarget symptom across multiple conditions, then a moreglobal indication may be sought.ADHD: A sizable fraction of children with ADHD

have significant problems with aggressive behavior. In thelarge (N = 579) multisite Multimodal Treatment Studyof Children with ADHD (MTA Cooperative Group,1999), four groups (optimal medication, optimalbehavioral therapy, both combined, and usual commu-nity care) were compared. After 14 months of optimalmedication treatment for children in themedication-onlyand combined treatment groups, 44% of 267 childrenwith initial aggression remained significantly symptom-atic. The continuing high levels of IA in this relativelyunresponsive subgroup (constituting 26% of the totalMTA sample) may have otherwise warranted some formof augmentation treatment, had the study design allowedfor this option. (See supporting data tables prepared bycoauthor P.J. via the Article Plus feature.) Although theMTA study design did not address the potential merits of

Fig. 2 Percentage of children meeting criteria for no, mildYmoderate, orsevere impulsive aggression construct. BP II = bipolar disorder II; BP NOS =bipolar disorder not otherwise specified; CBCL = Child Behavior Checklist;Disruptive Beh D/O = disruptive behavior disorders

JENSEN ET AL.

6 J . AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 46:3, MARCH 2007

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

additional pharmacological treatments, other researchsuggests that IA when comorbid with ADHD mayrespond to augmentation treatments, compared toplacebo (Aman et al., 2004).

Of interest are the treatment recommendations for theuse of atypical antipsychotic drugs in aggressive youths(TRAAY), a set of consensus recommendations adoptedby New York State and Columbia University inpartnership with leading investigators nationwide(Pappadopulos et al, 2003). Key aspects of the TRAAYalgorithm are to treat the primary disorder with first-linetreatments first (not aggression per se), to use mono-therapy whenever possible, to use psychosocial andbehavioral treatments for aggression, and, if/when all ofthese initial steps fail, to move on to concurrent use of anatypical antipsychotic drug to manage aggression.

Autism: In addition to the formal criteria for autism(onset before age 3, marked impairment in socialinteraction, restricted/repetitive patterns of behavior),DSM-IV indicates that other symptoms often becomethe target of treatment, such as aggressiveness, self-injurious behavior, severe temper tantrums, hyperactiv-ity, and impulsivity. The class of drugs most studied forreducing these symptoms are the antipsychotics, inparticular haloperidol (e.g., Campbell et al., 1978) andrisperidone (Findling and McNamara, 2004; Snyderet al., 2002). Studies of these agents suggest that theyboth reduce symptoms associated with autism (includ-ing aggression), however, it is unclear whether earlystudies (haloperidol) and later studies (risperidone) arecomparable, given substantial differences in studydesign, including patient selection and aggressionassessment methods. Thus, in a recent study of childrenwith autism (ages 5Y17 years) who also met a criterionfor disruptive behavior, risperidone treatment yieldedsubstantial reductions in aggression compared withplacebo (Research Units on Pediatric Psychopharma-cology Autism Network, 2002). In contrast, in studiesof children with autism (ages 2Y7 years) not selected foraggression, the effect of haloperidol on aggressivesymptoms appeared more variable (Anderson et al.,1984, 1989). An early meta-analysis of available studiesof haloperidol did generally indicate some improve-ment in anger, uncooperativeness, and aggression, aswell as hyperactivity and speech deviance in the druggroup (Locascio et al., 1991).

Bipolar Disorder: Aggression is frequently reportedin pediatric bipolar disorder. As part of the first phase of a

multiphase prospective clinical trial, 139 youths ages 5 to17 years with bipolar disorder were treated withcombination lithium and divalproex sodium (Findlinget al., 2003, 2005). Impulsive aggression was measuredwith the Young Mania Rating Scale (Young et al., 1978)and the Parent General Behavior Inventory (Youngstromet al., 2001). Almost all of the participants showed someIA, but a substantial subgroup did not (Fig. 2). Althoughaggression explained only a small part of the variance inclinician-rated severity of manic symptoms, whencompared with other symptoms of mania, aggressionappeared to be less responsive to treatment and a keyfactor inmediating an inadequate therapeutic response inbipolar patients. Moreover, IA severity substantiallypredicted withdrawal from treatment in this study,again suggesting the importance of clearly separatingimpulsive aggressive symptoms from other symptoms ofthe primary disorder (see supporting data tables preparedby coauthor E.Y. via the Article Plus feature).The essential features of IA can and should be

operationalized by clear clinical and research criteria,ascertained by careful clinical interview.It is essential to define the inclusion and exclusion

criteria and the clinical endpoints in clinical trials.Clearly naming, defining, and identifying the clinicalentity (including distinguishing it from other clinicalentities) is needed to gain FDA approval of a newdrug claim for a symptom or disorder. IA assessmentsshould characterize the history/nature/type of specificimpulsive aggressive acts and their maladaptivenature, frequency, severity, duration, cross-situation-ality, triggers, contexts, consequences, presence ofsignificant impairment, age of onset, and so forth.Such assessments must gather information about thechild_s aggression within already agreed-on and well-operationalized diagnostic categories (ADHD, bipolar,autism, PTSD), accompanied by careful specification ofaggressive symptoms above a certain severity and/orimpairment threshold. This information should beobtained by structured or semistructured interviews,both for ascertaining the specific nature of aggressionand primary diagnoses. Such a semistructured interviewis under development (modified from Barratt et al.,1999) and should define the nature, number, type,duration, and severity of specific impulsive aggressiveacts in the context of the lack of major triggers, theircross-situationality contexts, and other factors (e.g., seeSteiner, 2005a, b).

IMPULSIVE AGGRESSION IN CHILD PSYCHIATRY

7J . AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 46:3, MARCH 2007

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

To develop appropriate participant groups for studyentry, new assessment tools are desirable but not essentialfor new pharmacological studies to proceed in pursuit of apotential indication.To determine the appropriateness of patients for

referral to clinical trials in this therapeutic area,conference participants began with the assumptionthat most subjects would be referred from othertreatment settings, generally because of the failure orlimited effectiveness of treatments for conditions suchas ADHD, bipolar disorder, or autism. Thus, it wouldbe necessary to ensure that treatments for the Axis IDSM-IV disorder were adequate and that the investi-gators assess residual symptoms of IA before enteringsuch children into the trial. Thus, claims for treatmentindications of IA should be restricted to the disorderstudied. By focusing on patients who have beencarefully screened, evaluated, and determined to havepersistent IA despite effective treatment of the coreDSM-IV Axis I disorder, positive results of treatmentof IA are less likely to raise a concern aboutpseudospecificity.Age and gender also need to be considered because

younger children (and boys) are more prone toirritability, temper tantrums, and other disruptivebehaviors (Dodge, 1991; Dodge et al., 1997), anddevelopmental differences should not be confused withnon-normative, high levels of maladaptive IA in olderpatients. For example, a 4-year-old hitting anotherchild must be considered differently from the samebehavior in a 14-year-old. Likewise, punching a siblingmay be considered differently from striking a teacher.Studies should be restricted to children older than 5years, except in the instance of PDD or autism, inwhich children older than 3 years may be included.Conference participants agreed on the importance ofassessing (and/or ruling out, depending on the studyquestions) the presence of psychosis, mental retarda-tion, PDD, speech/language problems, nonverballearning disabilities, substance abuse/dependence, leadintoxication, and severe traumatic brain injury becauseall of these factors have been found to be linked toincreased aggression. The patient_s IQ and familyhistory, the history of any physical or psychologicaltrauma or abuse, and an assessment of overall physicalhealth should also be obtained. Finally, extendingtreatment studies to different cultural/ethnic groupsraises concerns that aggression measures may not

necessarily be interpreted in the same ways by each ofthe populations being studied. Such considerationsrequire possible adaptation of questions that may bemisinterpreted or misunderstood by diverse respon-dents. Field testing of measures, including the use offocus groups, should be considered to ensure item andquestionnaire reliability and validity across cultural/ethnic groups.Three types of assessments of IA symptoms may

constitute a menu from which to select and todetermine study entry: (1) trait measures, the under-lying propensity to become physically or verballyaggressive with only minor provocations or frustrations;(2) characterizing/assessing recent impulsive aggressiveacts; and (3) laboratory-based measures.Currently available measures that tap the trait of IA

include the Buss-Durkee Hostility Inventory (Buss andDurkee, 1957), the Life History of Aggression (Coccaroet al., 1997), the proactive versus reactive aggressionscale extracted from the Achenbach instruments(Dodge and Coie, 1987), the State-Trait AngerExpression Inventory (Reyes et al., 2003), the VerbalAggression and Provoked Physical Aggression subscalesfrom the Children_s Aggression Rating Scale (Halperinet al., 2002), and aggression subscales from the ChildBehavior Checklist (Achenbach, 1991; Achenbach andRescorla, 2001). Other assessments that may be usefulinclude the Conduct Problem subscale from theNisonger Child Behavior Rating Form, and theIrritability subscale from the Aberrant Behavior Check-list. Parents will usually be the primary reporter,although in some instances youth reports have provenuseful (Nickel et al., 2005, 2006). However, ifpervasiveness of aggression is important, then teacherinformants will be required. Depending in part on thepopulation and disorder studied, child self-reports alsomay be included.Retrospective recall of information about the

frequency or intensity of recent aggressive-impulsiveacts or episodes may be one approach to characterizing/assessing recent impulsive aggressive episodes. Depend-ing on the population and the aggression severity, thequestionnaire or interview should ask respondents torecall episodes over the past 3 months, 4 weeks, 2weeks, and past week. The Modified Overt AggressionScale, which characterizes specific aggressive acts, andthe Overt Aggression Scale-Modified of Coccaro et al.(1991, 1997) (see Malone et al., 2004 for use in

JENSEN ET AL.

8 J . AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 46:3, MARCH 2007

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

children and adolescents) may be adapted for inter-viewing parents and children separately. Clearly, moredevelopment and study of possible IA measures inchildren and youths are needed.

Response inhibition, consequence sensitivity, startleresponse, and psychophysiological measures (e.g.,autonomic responsivity) are useful laboratory-basedmeasures that tap cognitive, emotional, and physiolo-gical domains relevant to aggression and/or impulsivity.By themselves, these measures are not sufficient todefine inclusion criteria (or primary endpoints), butthey may provide useful adjunctive information inproof of concept studies and may add some objectiveface validity to self-reports or parental reports.Laboratory measures, although expensive, may havemerits in selected studies attempting to determine theimpact of treatments on underlying brain processes andmechanisms of drug action.

Treatment response should be evaluated by a minimumof two complementary strategies: measurement of specificaggressive acts and measurement of aggressive/impulsivetraits (and, in some studies, data from laboratory-basedassessments of cognitive, emotional, and psychophysiologicfunctions associated with impulsivity or aggression).

Although each of the areas of assessment described inthe previous section are also relevant to a considerationof treatment efficacy, a well-designed clinical trialrequires the definition of a validated primary endpoint,as well as consideration of relevant and informativesecondary endpoints. Conference participants recom-mended that at a minimum, these two complementarystrategies must be evaluated to assess treatmentresponses (as primary and secondary endpoints), butthe primary endpoint, in general, should be a change inthe severity, frequency, and course of impulsiveaggressive acts during treatment compared to baseline.Diary data should address counts, frequency, andseverity of aggressive behaviors, whether the aggressionis cross-situational, and the target of the aggressive act.Such data may be collected by diary or PDA and shouldbe keyed to some scale, such as the Modified OvertAggression Scale, to allow for data gathering efficiencyand concurrent validation.

IED may be a promising diagnostic category in youths,but there are few data available to support or require itsuse in lieu of the construct of IA.

IED is a categorical expression of IA and is moreprevalent than previously thought, with 7% and 4%

lifetime and 12-month prevalence rates, respectively(Coccaro et al., 2004; Kessler et al., 2006). Relatedly, inanother study recently examining IED in a sample of1,300 adult psychiatric outpatients, Coccaro et al.(2005) found that among with adult patients with IED(6.3% of the clinical sample), 30% had IED onsetduring their preteen years and about 75% had onset bythe end of their teenage years. Although little researchon IED in youths has been done to date, given theapparent usefulness of the construct in studies of adultswith aggression, further studies of the applicability ofthe IED diagnostic category in youths are warranted.Such studies should focus on the question of IEDcomorbidity in children with other, currently acceptedDSM-IV diagnoses such as bipolar, ADHD, PTSD,and autism/PDD.Treatment trials should aim to detect a signal of the

antiaggressive activity of drug X over some comparator(placebo or another drug).A major issue in designing IA treatment trials in a

cohort of patients with a specificDSM-IVAxis I disorderrelates to the state of FDA-approved treatments for Axis Idisorders in child and adolescent psychiatry. Althoughthere are FDA-approved drugs for ADHD, it may becomorbid with disruptive disorders (e.g., oppositionaldefiant disorder, conduct disorder, conduct disorder nototherwise specified) for which there are no approvedmedications. No drugs have been approved for treatingautism, and no drugs have yet demonstrated efficacy inregistration trials for the treatment of bipolar disorder inchildren and adolescents. A previous consensus con-ference (Carlson et al., 2003) should serve as a guidepostin the design of studies of bipolar disorder in childrenand adolescents. Where drug treatments have beenapproved for the treatment of some other disorders(outside of child and adolescent psychiatry), the FDAhasapproved other drugs for specific aspects of thesedisorders that are not well addressed by standardtherapies (e.g., agitation and suicidality in schizophrenia)and in other cases has endorsed development programsthat target specific symptoms in certain disorders (e.g.,psychosis in Alzheimer_s disease, cognitive impairmentin schizophrenia).Thus, several design options are possible: (1) a two-

arm add-on clinical trial design of patients with IA co-occurring with ADHD (and possibly also oppositionaldefiant disorder and/or conduct disorder), (2) a three- orfour-arm comparison of monotherapy versus placebo

IMPULSIVE AGGRESSION IN CHILD PSYCHIATRY

9J . AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 46:3, MARCH 2007

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

and/or combination therapy in bipolar disorder, and (3) atwo-arm monotherapy (versus placebo) study design inIA in autistic patients (T2 Table 2 for various design andendpoint options recommended). Whether as an add-ontherapy or as a monotherapy in patients with IA, studydesigns should be powered to provide documentation ofthe antiaggressive efficacy of the Bserenic[ (antiaggressivebut not sedating) treatment.The easiest situation to address is for a disorder that

has standard treatments approved, but where thesetreatments do not adequately address IA as an aspect ofthat disorder (e.g., ADHD). An add-on design may beoptimal for such situations. Another relatively straight-forward design situation may concern a disorder thathas no treatments for the primary disorder, but whereIA is an aspect of the disorder that needs management(e.g., IA in autism). In such situations, testing a serenicversus placebo may be the most straightforward designoption.Other situations may not be so straightforward. For

example, no drugs are yet specifically approved forpediatric bipolar disorder, although there are adultmedications that may be expected to offer benefit inchildren. In this situation, it may be difficult to evaluatea trial testing a serenic versus placebo, with IA as the

primary endpoint. A three-arm design may be a viableoption, but studies involving comparisons betweendrugs often raise questions regarding fairness of thecomparisons.Given that IA tends to be chronic, the duration of

FDA-approvable treatment studies were considered,with recommendations as follows:

• For initial efficacy: 6 weeks. Efficacy should beidentifiable after 3 to 4 weeks of steady-statetreatment (with studies lasting up to 6 weeks todetect a signal), but this will depend on the specificprimary disorder and the particular agent. Previouscontrolled studies have established that efficacy mayeven be noted after 1 week, at least in studies ofatypical agents (Schur et al., 2003).

• For long-term efficacy: 6 months of maintenancetherapy. This length is not altogether arbitrary, giventhe fact that previous studies have shown thatbenefits may persist up to and even beyond thisperiod of time Findling et al., 2004; Reyes et al.,2006; Schur et al., 2003).

• For safety: 1 year. Follow-up studies of side effects oftreated aggression in children and youths have beenconducted, appear feasible, and suggest that some

TABLE 2Recommended Designs for IA Clinical Trials

Disordera Design Options Endpoint Options

ADHD Two-arm study: serenic vs. placeboas add-on treatment inpatients receiving optimalapproved treatment for ADHD; OR

Primary: IA symptoms

Three-arm study: stimulant + placebovs. serenic + placebo vs. stimulant + serenic;OR stimulant vs. serenic vs. placebo

Secondary: IA trait measures,or residual symptomsof ADHD

Autism Two-arm study: serenic vs. placebo Primary: IA symptomsSecondary: IA trait measuresor autistic symptoms

Bipolar disorder Two-arm study: serenic vs. placebo Primary: IA symptomsThree-arm study: mood stabilizingdrug vs. serenic vs. placebo

Secondary: IA trait measuresor other core symptomsof bipolar disorder in children

Four-arm study: same as aboveplus combination of moodstabilizing drug + serenic

Note: IA = impulsive aggression; ADHD = attention-deficit/hyperactivity disorder.a Design recommendations for other conditions, such as pervasive developmental disorder, personality disorders, posttraumatic stress

syndrome, oppositional defiant disorder, and conduct disorder were not addressed by the conference participants, given the controversial natureof some of these conditions (e.g., conduct disorder) or the lack of FDA-approved medication indications.

JENSEN ET AL.

10 J . AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 46:3, MARCH 2007

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

initial side effects may dissipate over this period oftime (Findling et al., 2004).

After efficacy/tolerability has been proven, long-termeffectiveness should be further documented with aplacebo-controlled discontinuation trial, after firststabilizing for >6 months of extended treatment,followed by randomly assigning subjects to placebo orcontinued therapy.

Special ethical considerations apply and must beaddressed in clinical studies of IA in children and youths.

• Children in foster care or the juvenile justice systemmay be at special risk for inappropriate inclusion instudies of aggression. The possibility of such studiesshould not be excluded, however, because the mostseverely disturbed children with aggression are oftenfound in these settings. At this juncture, priorityshould be given to studies limited to children withparents or legal guardians who can provide somehistory and informed consent and help monitor thechildren throughout the study. Nonetheless, childrenin the juvenile justice system have higher rates of IA,and studies of this population are clearly warrantedwhen the necessary ethical protections are in place.

• In all studies, study developers must be cognizant ofhow the local community views aggression andwhether patients, their families, and others in thecommunity from which subjects are drawn acceptthis as a disease or as a symptom necessitating aclinical intervention. In particular, studies includingcultural/ethnic groups who have a history of beingharmed by research participation and are distrustfulof researchers would benefit from careful communityconsultation through all phases of the researchprocess. Researchers should consider the use ofmechanisms such as a community advisory boardto ensure the integrity of this process. Because thesestudies are likely to involve multiple institutions andresearch sites, researchers and study sponsors shouldconsider including such community consultationand advisory functions at the local, institutional, andcoordinating center levels.

• The process of clinical trials must be made astransparent as possible without compromising theintegrity of the research. It is important to make surethat parents, youths, and appropriate communityrepresentatives understand the study, its purposes,and its eventual possible outcomes and that informa-

tion is communicated to them on an ongoing basis inunderstandable language. These are basic buildingblocks for trust and for ensuring that the studyachieves its ultimate goals and is not misconstrued byothers in the media or general public.

• Good information must be provided about potentialadverse effects before, during, and after the trial hasbeen completed, including new data collected in thetrial. In the wake of the selective serotonin reuptakeinhibitor black box warning for adolescents and thedemand by leading journals that both positive andnegative trials be reported, full disclosure of adverseeffects is essential. If such data remain unpublished,pediatric prescribers may continue to prescribe off-label, unaware of specific side effects in children.Encouraging open disclosure and publication of alltrials would keep the field more informed. Inaddition, trials should include systematic, prospec-tive collection of adverse events information usingstate-of-the-art assessments to prevent the interpreta-tion difficulties that can occur in post hoc analyses(e.g. selective serotonin reuptake inhibitor trials).The more carefully that such data are collected, themore valid any safety assessments and resultantpatient-oriented information will be.

• Study designs must ensure that medical, mental, andenvironmental factors (biological, psychological, andsocial) that may be causing or contributing to theaggression are identified. When identified, if suchfactors may be otherwise effectively remediated, thesesteps should be attempted before study entry. Youthswith IA caused by conditions that have been shown tobe curable, treatable by other means, or attributable tospecific causes other than the specific type of IA beingstudied should be excluded from medication studiesuntil such factors have been satisfactorily addressedand the youths have been shown to have persistent highlevels of maladaptive IA.

• Ethical considerations may warrant that where anapproved treatment exists for a disorder (e.g.,ADHD), add-on studies may be the preferred clinicaltrials design (i.e., testing the study medication versusplacebo in children who are also being optimallytreated for their preexisting DSM condition) ratherthan to deprive them of a medication or othertreatment that is providing at least partial benefit.NIH-supported clinical research has demonstratedthat a variety of environmental and psychosocial

IMPULSIVE AGGRESSION IN CHILD PSYCHIATRY

11J . AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 46:3, MARCH 2007

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

treatments can be effective for children with severeaggression. For example, multidimensional treat-ment foster care (MTFC) targets youths placed ingroup homes or training schools because of seriousand chronic delinquency and/or aggression. Fosterparents are contacted daily and supported continu-ously, and youths receive family therapy, skillstraining, individual treatment, and psychiatric con-sultation. Trials of MTFC versus group care haveshown that measures of criminality and violence weremarkedly lower in MTFC versus group care youths(Chamberlain and Reid, 1998AQ1 ; Eddy et al., 2004).Although clinical trials examining the efficacy ofMTFC in combination with specific pharmaceuticalinterventions have yet to be done, where demon-strated effective psychosocial interventions exist,medication studies should be considered for add-onstudy designs (e.g., with medication or medicationsadded to a possibly efficacious psychotherapyintervention). Thus, trial designers may considerwhether a standard program of anger management orbehavioral therapy should be made available before,during, or throughout the efficacy trial. It may beduring the midst of such trials (usually characterizedby intensive monitoring and support, whether theparticipant is taking placebo or active drug) that newbehavioral skills may be more easily learned andapplied so that children and families have thenecessary tools to cope more effectively aftermedication withdrawal and study cessation.

• BSafety nets[ for children and families participatingin such trials are essential. Because some childrenmay show deterioration during a trial, exit rules andimmediate access to a medical setting must beavailable if clinical worsening occurs to help ensurethat the child is appropriately treated rather thanending up in the juvenile justice system. Parents havethe same concerns in clinical trials as they do outsidethem: BWho_s going to treat my child?[ It isenormously challenging for families to deal withthese youngsters, and they must have a clear path ofoptions to pursue when problems arise.

DISCUSSION

The central finding of conference participants was thatIA is in fact a sufficiently similar construct acrossdiagnostic categories, such that studies across several

disorders can and should be conducted. Such studies, ifappropriately designed, are interpretable and appro-priate to pursue medication indications. However,because the likelihood of detecting a signal for anefficacious treatment is best achieved when clinical trialsfocus on a single and distinct underlying DSM-defineddisorder, given the current state of knowledge, IAshould be studied principally within well-defineddiagnostic groups, such as ADHD, autism, and bipolardisorder. It is understood that the safety/efficacy claimshould not be extrapolated beyond the disorder that isstudied to extend it to IA under all circumstances.Although a gold standard for measuring IA does notexist and assessment methods are in need of furtherrefinement, consensus was reached that current assess-ment instruments are adequate to conduct IA treatmentstudies. Additional studies of valid and reliable IAmeasures are urgently needed, however.

Clinical Implications

An increased appreciation of the differences betweenIA and proactive/planned aggression and the applica-tion of these distinctions in clinical, research, andmedication development programs are likely to have amajor impact on future treatment planning, publicpolicy, and prevention programs. Clinicians andresearchers alike should continue to explore and refinethese distinctions to determine their full implicationsfor our understanding of psychopathology, pathophy-siology, treatment response, clinical course, andoutcome.

Limitations

It should be noted that the final recommendations inthis report were derived by a combination of datareanalyses, literature reviews, and expert consensusmethods and are subject to human error or mistakes ininterpretation. Thus, continued research of the meritsof the constructs of IA and proactive/planned aggres-sion is needed, as is an ongoing examination of theimpact of these recommendations on future studies ofaggression in children and youths.

Disclosure: Dr. Jensen receives investigator-initiated grants fromMcNeil and unrestricted educational grants from Pfizer, Eli Lilly,and McNeil; participates on the speakers_ bureau of CMED, UCBPharma, PsychCME, CME Outfitters, and the Neuroscience EducationInstitute; and consults with Best Practice, Inc., Janssen, Novartis, andUCB Pharma. Dr. Youngstrom receives grant support from Abbott,AstraZeneca, and BMS and consults with Best Practices, Eli Lilly, and

JENSEN ET AL.

12 J . AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 46:3, MARCH 2007

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

Otsuka. Dr. Steiner participates on the speakers_ bureau of Abbott,Janssen, and Eli Lilly; serves as an advisor/consultant for Abbott,Janssen, and Otsuka; receives unrestricted educational grants fromAbbott, Forest, GlaxoSmithKline, Janssen, McNeil, Novartis, Pfizer,Solvay, Wyeth-Ayerst, Ortho-McNeil, and Eli Lilly; and has receivedresearch grants from NIJ, TCWFAQ2 , Eucalyptus, Abbott, Janssen, andPfizer. Dr. Meyer is CEO of Best Practice Project Management, Inc., aconsulting and project management company specializing in the centralnervous system; his company receives support from Abbott, AstraZeneca,BMS, Cephalon, DAS, GlaxoSmithKline, Janssen, Johnson & JohnsonResearch, Eli Lilly, Memory, NADI, the National Institute of MentalHealth (UCLA), Novartis, Ovation, Pfizer, Sanofi, Sepracor, Solvay,Star Scientific, and Takeda. Dr. Malone receives grant support fromBristol-Myers Squibb, Eli Lilly, Pfizer, and Shire-Richwood. Dr.Carlson receives grant support from Abbott, Eli Lilly, Janssen, andOtsuka and has consulted to McNeil, Johnson & Johnson, Pfizer, andNovartis. Dr. Findling receives grant support from many pharmaceu-tical companies including Abbott, AstraZeneca, BMS, Celltech UCB,Forest, GlaxoSmithKline, Johnson & Johnson, Eli Lilly, Novartis,Noven, Otsuka, Pfizer, Shire, Solvay, and Wyeth; consults to the samecompanies; and participates on the speakers_ bureau of Shire-Richwood.Dr. Aman participates on advisory boards for Janssen, Novartis, EliLilly, Bristol-Myers Squibb, and Abbott and has investigator-initiatedcontracts with Janssen and Eli Lilly. Dr. Coccaro has been a consultantfor Eli Lilly, Abbott, and AstraZeneca and has had investigator-initiated contracts with Eli Lilly and Abbott. Dr. Ferris has a financialinterest in Serenix Pharmaceuticals and Rexahn Pharmaceuticals,consulting on the development of drugs for the treatment of impulsivityand violence. Dr. Novins consults with Prescription Solutions, Inc.,regarding prescribing guidelines for medications prescribed to childrenand adolescents for the treatment of psychiatric disorders and withMcNeil regarding their initiative to reduce disparities in children_smental health research in minority communities. The other authorshave no financial relationships to disclose.

REFERENCES

AACAP (1997), Practice parameters for the assessment and treatment ofchildren and adolescents with conduct disorder. J Am Acad Child AdolescPsychiatry 36:122SY139S

Achenbach TM (1991), Manual for the Child Behavior Checklist 4Y18 and1991 Profile. Burlington: University of Vermont, Department ofPsychiatry

Achenbach TM, Rescorla LA (2001), Manual for the ASEBA School-AgeForms and Profiles. Burlington: University of Vermont, Department ofPsychiatry

Aman MG, Binder C, Turgay A (2004), Risperidone effects in the presence/absence of psychostimulant medicine in children with ADHD, otherdisruptive behavior disorders, and subaverage IQ. J Child AdolescPsychopharmacol 4:243Y254

Anderson LT, Campbell M, Grega DM, Perry R, Small AM, Green WH(1984), Haloperidol in the treatment of infantile autism: effects onlearning and behavioral symptoms. Am J Psychiatry 141:1195Y1202

Anderson LT, Campbell M, Adams P, Small AM, Perry R, Shell J (1989),The effects of haloperidol on discrimination learning and behavioralsymptoms in autistic children. J Autism Dev Disord 19:227Y239

Barratt ES, Stanford MS, Dowdy L, Liebman MJ, Kent TA (1999),Impulsive and premeditated aggression: a factor analysis of self-reportedacts. Psychiatry Res 86:163Y173

Best M, Williams JM, Coccaro EF (2002), Evidence for a dysfunctionalprefrontal circuit in patients with an impulsive aggressive disorder. ProcNatl Acad Sci U S A 99:8448Y8453

Blair RJR (2004), The roles of orbital frontal cortex in the modulation ofantisocial behavior. Brain Cogn 55:198Y208

Buss AH, Durkee A (1957), An inventory for assessing different kinds ofhostility. J Consult Psychol 21:343Y348

Campbell M, Anderson LT, Meier M et al. (1978), A comparison ofhaloperidol and behavior therapy and their interaction in autisticchildren. J Am Acad Child Psychiatry 17:640Y655

Carlson GA, Jensen PS, Findling RL et al. (2003), Methodological issuesand controversies in clinical trials with child and adolescent patients withbipolar disorder: report of a consensus conference. J Child AdolescPsychopharmacol 13:13Y27

Chamberlain P, Reid JB (1998), Comparison of two community alternativesto incarceration for chronic juvenile offenders. J Cons Clin Psychol66:624Y633

Coccaro EF, Berman ME, Kavoussi RJ (1997), Assessment of life history ofaggression: development and psychometric characteristics. PsychiatryResearch 73:147Y157

Coccaro EF, Harvey PD, Kupsaw-Lawrence E, Herbert JL, Bernstein DP(1991), Development of neuropharmacologically based behavioralassessments of impulsive aggressive behavior. J Neuropsychiatry ClinNeurosci 3:S44YS51

Coccaro EF, Kavoussi RJ (1997), Fluoxetine and impulsive aggressivebehavior in personality disordered subjects. Arch Gen Psychiatry54:1081Y1088

Coccaro EF, Posternak MA, ZimmermanM (2005), Prevalence and featuresof intermittent explosive disorder in a clinical setting. J Clin Psychiatry66:1221Y1227

Coccaro EF, Schmidt CA, Samuels JF, Nestadt G (2004), Lifetime and 1-month prevalence rates of intermittent explosive disorder in acommunity sample. J Clin Psychiatry 65:820Y824

Connor DF (2002), Aggression and Antisocial Behavior in Children andAdolescents: Research and Treatment. New York: The Guilford Press

Connor DF, Steingard RJ, Anderson JJ, Cunningham JA, Melloni RH Jr(2004), Proactive and reactive aggression in referred children andadolescents. Am J Orthopsychiatry 74:129Y136

Dodge KA (1991), The structure and function of reactive and proactiveaggression. In: The Development and Treatment of Childhood Aggression,Pepler DJ, Rubin KH, eds. Hillsdale, NJ: Lawrence Erlbaum Associates,pp 201Y218

Dodge KA, Coie JD (1987), Social-information-processing factors inreactive and proactive aggression in children_s peer groups. J Pers SocPsychol 53:1146Y1158

Dodge KA, Lochman JE, Harnish JD, Bates JE, Pettit GS (1997), Reactiveand proactive aggression in school children and psychiatrically impairedchronically assaultive youths. J Abnorm Psychol 10:637Y651

Donovan SJ, Nunes EV, Stewart JW et al. (2003), Outer-directedirritability: a distinct mood syndrome in explosive youth with adisruptive behavior disorder? J Clin Psychiatry 64:698Y701

Eddy JM, Whaley RB, Chamberlain P (2004), The prevention of violentbehavior by chronic and serious male juvenile offenders: a 2-year follow-up of a randomized clinical trial. J Emot Behav Disord 12:2Y8

Ferris CF, DeVries GJ (1997), Ethological models for examining theneurobiology of aggressive and affiliative behaviors. In: Handbook ofAntisocial Behavior, Stoff D, Breiling J, Maser JD, eds. New York: Wiley,pp 255Y268

Findling RL, McNamara NK (2004), Atypical antipsychotics in thetreatment of children and adolescents: clinical applications. J ClinPsychiatry 65(suppl 6):30Y44

Findling RL, McNamara NK, Gracious BL et al. (2003), Combinationlithium and divalproex in pediatric bipolarity. J Am Acad Child AdolescPsychiatry 42:895Y901

Findling RL, Aman MG, Eerdekens M, Derivan A, Lyons B (2004), Long-term, open-label study of risperidone in children with severe disruptivebehaviors and below-average IQ. Am J Psychiatry 161:677Y684

Findling RL, McNamara NK, Youngstrom EA et al. (2005), A double blind18-month trial of lithium vs. divalproex maintenance treatment inchildren and adolescents with bipolar disorder. J Am Acad Child AdolescPsychiatry 44:409Y417

Halperin JM, McKay KE, Newcorn JH (2002), Development, reliabilityand validity of the Children_s Aggression ScaleYParent Version. J AmAcad Child Adolesc Psychiatry 41:245Y252

IMPULSIVE AGGRESSION IN CHILD PSYCHIATRY

13J . AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 46:3, MARCH 2007

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

Hsieh DK, Kirk SA (2003), The effect of social context on psychiatrists_judgments of adolescent antisocial behavior. J Child Psychol Psychiatry44:877Y887

Kazdin AE, Bass D, Siegel T, Thomas C (1989), Cognitive-behavioraltherapy and relationship therapy in the treatment of children referred forantisocial behavior. J Consult Clin Psychol 57:522Y535

Kazdin AE, Siegel TC, Bass D (1992), Cognitive problem-solving skillstraining and parent management training in the treatment of antisocialbehavior in children. J Consult Clin Psychol 60:733Y747

Kellam SG, Ling X, Merisca R, Brown CH, Ialongo N (1998), The effect ofthe level of aggression in the first grade classroom on the course andmalleability of aggressive behavior into middle school. Dev Psychopathol10:165Y185

Kessler RC, Coccaro EF, FavaM, Jaeger S, Walters E (2006), The prevalenceand correlates of DSM-IV intermittent explosive disorder in theNational Comorbidity Survey Replication. Arch Gen Psychiatry63:669Y678

Locascio JJ, Malone RP, Small AM et al. (1991), Factors related tohaloperidol response and dyskinesias in autistic children. Psychopharma-col Bull 27:119Y126

Lochman JE, Curry JF (1986), Effects of social problem-solving training andself-instruction training with aggressive boys. J Clin Child Psychol15:159Y164

Lochman JE, Lampron LB (1988), Cognitive-behavioral interventions foraggressive boys: seven months follow-up effects. J Child AdolescPsychotherapy 5:15Y23

Malone RP, Delaney MA, Gifford C (2004), Parent/child agreement onmeasures of aggression. Presented at the 51st Annual Meeting of theAmerican Academy of Child and Adolescent Psychiatry, Washington,DC, October 19Y24

Malone RP, Luebbert JF, Delaney MA et al. (1997), Nonpharmacologicalresponse in hospitalized children with conduct disorder. J Am Acad ChildAdolesc Psychiatry 36:242Y247

Malone RP, Luebbert J, Pena-Ariet M, Biesecker K, Delaney MA (1994),The Overt Aggression Scale in a study of lithium in aggressive conductdisorder. Psychopharmacol Bull 30:215Y218

McClure EB, Pope K, Hoberman AJ, Pine DS, Leibenluft E (2003), Facialexpression recognition in adolescents with mood and anxiety disorders.Am J Psychiatry 160:1172Y1174

MTA Cooperative Group (1999), A 14-month randomized clinical trial oftreatment strategies for attention deficit hyperactivity disorder. Arch GenPsychiatry 56:1073Y1086

Nickel MK, Krawczyk J, Nickel C et al. (2005), Anger, interpersonalrelationships, and health-related quality of life in bullying boys who aretreated with outpatient family therapy: a randomized, prospective,controlled trial with 1 year of follow-up. Pediatrics 116:e247Ye254

Nickel MK, Luley J, Krawczyk J et al. (2006), Bullying girlsVchanges afterbrief strategic family therapy: a randomized, prospective, controlled trialwith one-year follow-up. Psychother Psychosom 75:47Y55

Olfson M, Blanco C, Liu L, Moreno C, Laje G (2006), National trends inthe outpatient treatment of children and adolescents with antipsychoticdrugs. Arch Gen Psychiatry 63:679Y685

Pappadopulos E, Jensen PS, Schur SB et al. (2002), BReal world[ atypicalantipsychotic prescribing practices in public child and adolescentinpatient settings. Schizophr Bull 28:111Y121

Pappadopulos E, MacIntyre JC II, Sverd J et al. (2003), Treatmentrecommendations for the use of antipsychotics for aggressive youth(TRAAY). Part II. J Am Acad Child Adolesc Psychiatry 42:145Y161

Porter S (1998), Without conscience or without active conscience? Theetiology of psychopathy revisited. Aggress Viol Behav 1:179Y189

Poulin F, Boivin M (2000), Reactive and proactive aggression. Psychol Assess12:115Y122

Research Units on Pediatric Psychopharmacology Autism Network (2002),Risperidone in children with autism and serious behavioral problems. NEngl J Med 347:314Y321

Rey JM, Walter G, Plapp JM, Denshire E (2000), Family environment inattention deficit hyperactivity, oppositional defiant and conductdisorders. Aust N Z J Psychiatry 34:453Y457

Reyes LR, Meininger JC, Liehr P, Chan W, Mueller WH (2003), Anger inadolescents: sex, ethnicity, age differences, and psychometric properties.Nurs Res 52:2Y11

Reyes M, Buitelaar J, Toren P, Augustyns I, Eerdekens M (2006), Arandomized, double-blind, placebo-controlled study of risperidonemaintenance treatment in children and adolescents with disruptivebehavior disorders. Am J Psychiatry 163:402Y410

Schur SB, Sikich L, Findling RL et al. (2003), Treatment recommendationsfor the use of antipsychotics for aggressive youth (TRAAY). Part I: areview. J Am Acad Child Adolesc Psychiatry 42:132Y144

Snyder R, Turgay A, Aman M, Binder C, Fisman S, Carroll A, RisperidoneConduct Study Group (2002), Effects of risperidone on conduct anddisruptive behavior disorders in children with subaverage IQs. J Am AcadChild Adolesc Psychiatry 41:1026Y1036

Steiner H, Cauffman E (1998), Juvenile justice, delinquency and psychiatry.In: Child and Adolescent Psychiatric Clinics of North America, BerkowitzSJ, Adnopoz J, eds. Philadelphia: Saunders, pp 653Y672

Steiner H, Saxena K, Medic S, Plattner B, Delizonna L (2005a), Proactive/reactive aggression and psychopathology in high school students.Presented at the Annual Meeting of the American PsychiatricAssociation, Atlanta, May 24, 2005. Available at: http://www.psych.org/edu/other_res/lib_archives/archives/meetings/AMN/NR528.Accessed October 17, 2006

Steiner H, Delizonna L, Saxena K, Medic S, Plattner B, Haapanen R(2005b), Does the two factor model of aggression hold for incarcerateddelinquents? Presented at the Annual Meeting of the AmericanPsychiatric Association, Atlanta, May 24, 2005. Available at: http://www.psych.org/edu/other_res/lib_archives/archives/meetings/AMN/2005nra.cfm, NR529. Accessed October 17, 2006

Steiner H, Saxene K, Chang K (2003), Psychopharmacological strategies forthe treatment of aggression in youth. CNS Spectrums 8:298Y308

Vitaro F, Brendgen M, Tremblay RE (2002), Reactively and proactivelyaggressive children: antecedent and subsequent characteristics. J ChildPsychol Psychiatry 43:495Y505

Wakefield JC, Pottick KC, Kirk SA (2002), Should the DSM-IV diagnosticcriteria for conduct disorder consider social context? Am J Psychiatry159:380Y386

Young RC, Biggs JT, Ziegler VE, Meyer DA (1978), A rating scale formania: reliability, validity and sensitivity. Br J Psychiatry 133:429Y435

Youngstrom EA, Findling RL, Danielson CK, Calabrese JR (2001),Discriminative validity of parent report of hypomanic and depressivesymptoms on the general behavior inventory. Psychol Assess 13:267Y276

JENSEN ET AL.

14 J . AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 46:3, MARCH 2007