J. Child Psychol. Psychial. Vol. 38, No. 2. p. 161-178, 1997Cambridge University Press

© 1997 Association for Child Psychology and PsychiatryPrinted in Great Britain. All rights reserved

0021-9630/97 $15.00 + 0.00

Practitioner Review: Psychosocial Treatments for ConductDisorder in Children

Alan E. KazdinYale University, New Haven, U.S.A.

The present paper reviews promising treatments for conduct disorder among children andadolescents. The treatments include problem-solving skills training, parent managementtraining, functional family therapy and multisystemic therapy. For each treatment,conceptual underpinnings, characteristics and outcome evidence are highlighted. Limita-tions associated with these treatments (e.g. paucity of long-term follow-up evidence and ofevidence for the clinical significance of the change) are also presented. Broader issues thataffect treatment and clinical work with conduct-disordered youths are also addressed,including retaining cases in treatment, what treatments do not work, who responds well totreatment, comorbidity, the use of combined treatments and the need for new models oftreatment delivery.

Keywords: Conduct disorder, treatment, children and adolescents, therapy.

Abbreviations: ADHD: Attention Deficit/Hyperactivity Disorder; CD: Conduct Disorder;FFT: functional family therapy; MST: multisystemic therapy; ODD: OppositionalDefiant Disorder; PMT: parent management training; PSST: problem-solving skills train-ing.

Antisocial behaviors in children refer to a variety ofacts that reflect social rule violations and that are actionsagainst others. Behaviors such as fighting, lying andstealing are seen in varying degrees in most children overthe course of development. For present purposes, theterm conduct disorder will be used to refer to antisocialbehavior that is clinically significant and clearly beyondthe realm of "normal" functioning. The extent to whichantisocial behaviors are sufficiently severe to constituteconduct disorder depends on several characteristics ofthebehaviors including their frequency, intensity and chron-icity, whether they are isolated acts or part of a largersyndrome with other deviant behaviors, and whether theylead to significant impairment of the child as judged byparents, teachers or others.

Little in the way of effective treatment has beengenerated for conduct disorder. This is unfortunate inlight of the personal tragedy that conduct disorder canrepresent to children and their families and others whomay be victims of aggressive and antisocial acts. From asocial perspective, the absence of effective treatments isproblematic as well. Conduct disorder is one ofthe mostfrequent bases of clinical referral in child and adolescenttreatment services, has relatively poor long-term prog-nosis and is transmitted across generations (see Kazdin,

Requests for reprints to: Alan E. Kazdin, Department ofPsychology, Yale University, PO Box 208205, New Haven, CT06520-8205, U.S.A. (E-mail Kazdin@Minerva.cis.yale.edu).

1995b). Because children with conduct disorder oftentraverse multiple social services (e.g. special education,mental health, juvenile justice) the disorder is one of themost costly mental disorders in the United States(Robins, 1981).

There have been significant advances in treatment. Thepresent paper reviews research for four psychosocialtreatments that have shown considerable promise in thetreatment of conduct disorder in children and adoles-cents. ("Children" will be used to refer to both childrenand adolescents, unless a particular distinction is madebetween the two.) The treatments were selected becausethey have been carefully evaluated in controlled clinicaltrials. The paper describes and evaluates the under-pinnings, techniques and evidence on behalf of thesetreatments. Critical issues that are raised in providingtreatment to conduct disorder children and their familiesare also examined.

Overview of Characteristics of Conduct DisorderBefore discussing treatment of conduct disorder, it is

important to delineate the "problem" as it is oftenpresented clinically. From a treatment perspective, con-duct disorder represents a very broad domain involvingchild, parent, family and contextual conditions. Many ofthe factors that influence delivery and effectiveness oftreatment are not encompassed by the central diagnosticfeatures of the disorder. Next we will consider brieflysome sahent domains that are relevant to treatment.

161

162 A. E. KAZDIN

Central FeaturesThe overriding feature of conduct disorder is a per-

sistent pattern of behavior in which the rights of othersand age-appropriate social norms are violated. Isolatedacts of physical aggression, destruction of property,stealing and firesetting are sufficiently severe to warrantconcern and attention in their own right. Although thesebehaviors may occur in isolation, several of these arelikely to appear together as a constellation or syndromeand form the basis of a clinical diagnosis. For example, inthe Diagnostic and Statistical Manual of Mental Disorders(DSM-IV; American Psychiatric Association, t994), thediagnosis of Conduct Disorder (CD) is reached if thechild shows at least 3 of the 15 symptoms within in thepast 12 months, with at least 1 symptom evident withinthe past 6 months. The symptoms include: bullyingothers, initiating fights, using a weapon, being physicallycruel to others or to animals, stealing while confronting avictim, firesetting, destroying property, breaking intoothers' property, stealing items of nontrivial value,staying out late, running away, lying, deliberate firesettingand truancy.

It is important to retain the distinction betweenconduct disorder as a general pattern of behavior and thediagnosis of CD. The general pattern of conduct disorderbehavior has been studied extensively using varied popu-lations (e.g. clinical referrals and delinquent samples) anddefining criteria (Kazdin, 1995b). There is widespreadagreement and evidence that a constellation of antisocialbehaviors can be identified and has correlates related tochild, parent and family functioning. Moreover, anti-social behaviors included in the constellation extendbeyond those recognized in diagnosis (e.g. substanceabuse, associating with delinquent peers).

The Scope of DysfunctionIf one were to consider "only" the symptoms of

conduct disorder and the persistence of impairment, thechallenge of identifying effective treatments would begreat enough. However, the presenting characteristicsof children and their families usually raise a number ofother considerations that are central to treatment.Consider next the characteristics of children, parents,families and contexts that are associated with conductdisorder, as a backdrop for later comments on treat-ment.

Child characteristics. Children who meet criteria forCD are likely to meet criteria for other disorders as well.The coexistence of more than one disorder is referred toas comorbidity. In general, diagnoses involving disruptiveor externalizing behaviors (CD, Oppositional DefiantDisorder [ODD], and Attention Deficit/HyperactivityDisorder [ADHD]) often go together. In studies ofcommunity and clinic samples, a large percentage ofyouth with CD pr ADHD (e.g. 45-70%) also meetcriteria for the other disorder (e.g. Fergusson, Horwood& Lloyd, 1991; Offord, Boyle & Racine, 1991). Thecooccurrence of CD and ODD is common as well. Amongclinic-referred youth who meet criteria for CD, 84-96 %also meet concurrent diagnostic criteria for ODD (see

Hinshaw, Lahey & Hart, 1993).' CD is sometimescomorbid with anxiety disorders and depression (Hin-shaw et al., 1993; Walker et al., 1991).

Several other associated features of CD are relevant totreatment. For example, children with conduct disorderare also likely to show academic deficiencies, as reflectedin achievement level, grades, being left behind in school,early termination from school and deficiencies in specificskill areas such as reading. Youths with the disorder arelikely to evince poor interpersonal relations, as refiectedin diminished social skills in relation to peers and adultsand higher levels of peer rejection. Conduct disorderyouths also are likely to show a variety of cognitive andattributional processes. Deficits and distortions in cog-nitive problem-solving skills, attributions of hostile intentto others, and resentment and suspiciousness, illustrate afew cognitive features associated with conduct disorder.

Parent and family characteristics. Several parent andfamily characteristics are associated with conduct dis-order (see Kazdin, 1995b; Robins, 1991; Rutter & Giller,1983). Criminal behavior and alcoholism are two of thestronger and more consistently demonstrated parentalcharacteristics. Parent disciplinary practices and atti-tudes, especially harsh, lax, erratic and inconsistentdiscipline practices, often characterize the parents. Dys-functional relations are also evident, as refiected in lessacceptance of their children, less warmth, affection andemotional support, and less attachment, compared toparents of nonreferred youth. Less supportive and moredefensive communications among family members, lessparticipation in activities as a family and more cleardominance of one family member are also evident. Inaddition, unhappy marital relations, interpersonal con-flict and aggression characterize the parental relations ofantisocial children. Poor parental supervision and moni-toring of the child and knowledge of the child's where-abouts are also associated with conduct disorder.

Contextual conditions. Conduct disorder is associatedwith a variety of untoward living conditions such as largefamily size, overcrowding, poor housing, and dis-advantaged school settings (see Kazdin, 19951)). Many ofthe untoward conditions in which families live placestress on the parents or diminish their threshold forcoping with everyday stressors. The net effect can beevident in parent-child interaction in which parentsinadvertently engage in patterns that sustain or accelerateantisocial and aggressive interactions (e.g. Dumas &Wahler, 1983; Patterson, Capaldi & Bank, 1991).

Ouite often the child's dysfunction is embedded in alarger context that cannot be neglected in conceptualviews about the development, maintenance and course ofconduct disorder nor in the actual delivery of treatment.For example, at our outpatient clinical service (YaleChild Conduct Clinic), it is likely that a family referredfor treatment will experience a subset of these charac-teristics: financial hardship (unemployment, significantdebt, bankruptcy), untoward living conditions (danger-

^ In DSM-IV, if the child meets criteria for CD, ODD is notdiagnosed, because the former is likely to include manysymptoms of the latter. Yet, invoking and evaluating the criteriafor these diagnoses ignoring this consideration has been usefulin understanding the relation and overlap of these diagnoses.

TREATMENT OF CONDUCT DISORDER 163

ous neighborhood, small living quarters), transportationobstacles (no car or car in frequent repair, state providedtaxi service), psychiatric impairment of one ofthe parents,stress related to significant others (former spouses,boyfriends or girlfriends) and adversarial contact with anoutside agency (schools, youth services, courts). Conductdisorder is conceived as a dysfunction of children andadolescents. The accumulated evidence regarding thesymptom constellation, risk factors, and course overchildhood, adolescence and adulthood attests to theheuristic value of focusing on individual children. At thesame time, there is a child-parent-family-context gestaltthat includes multiple and reciprocal influences that aifecteach participant (child and parent) and the systems inwhich they operate (family, school) (Kazdin, 1993). Fortreatment to be effective, it is likely that multiple domainswill have to be addressed.

Promising Treatment ApproachesOverview: Criteria for Identifying Promising Treat-ments

Many different treatments have been applied to con-duct-disordered youths, including psychotherapy, phar-macotherapy, psychosurgery, home, school and com-munity-based programs, residential and hospital treat-ment, and social services (see Brandt & Zlotnick, 1988;Dumas, 1989; Kazdin, 1985; United States Congress,1991). Of the over 230 documented psychotherapiesavailable for children and adolescents (Kazdin, 1988), thevast majority have not been studied. Among those thathave, none has been shown to controvert conductdisorder and its long-term course. Many treatments mightseem conceptually justified as interventions for conductdisorder. Conduct disorder is a dysfunction with per-vasive features so that one can point to virtually anydomain (e.g. psychodynamics, family interaction pat-terns, cognitive deficiencies) and find aberrations, deficitsand deficiencies.

In our own work, we have relied on several criteria(please see Table 1) to identify and to select promisingtreatments among the array of available interventions.The initial criterion is that the treatment should havesome theoretical rationale that notes how the dysfunc-tion, in this case conduct disorder, comes about and thenhow treatment redresses the dysfunction. Specification ofthe mechanisms leading to conduct disorder and leadingto therapeutic change are required for this initial cri-terion.

The second criterion considers whether there is anybasic research to support the conceptuaUzation. Basicresearch in this context refers to studies that examineconduct problems and factors that lead to their onset,maintenance, exacerbation, amelioration or attenuation.An example would be studies of the family that dem-onstrate specific interaction patterns among parents andchildren that exacerbate aggression within the home(Patterson, Reid & Dishion, 1992). Such research wouldadvance considerably the conceptual view that positedthe significance of these patterns and provided a warrantfor treatments that are aimed at these interaction pat-terns.

The third criterion is whether there is any outcome

Table 1Criteria for Identifying Promising Treatments1. CONCEPTUALIZATION

Theoretical statement relating the mechanism(s) (e.g.intrapsychic, intrafamilial) to clinical dysfunction

2. BASIC RESEARCHEvidence showing that the mechanism can be assessed andrelates to dysfunction, independently of treatment outcomestudies

3. PRELIMINARY OUTCOME EVIDENCEEvidence in analogue or clinical research showing that theapproach leads to change on clinically relevant measures

4. PROCESS-OUTCOME CONNECTIONEvidence in outcome studies showing a relationshipbetween the change in processes alleged to be operative andclinical outcome

evidence that the treatment can effect change. In canvas-sing the literature, we tend to be very lenient for invokingthis criterion; we are interested in any demonstration(e.g. so-called open trials, studies with mildly disturbedcases). Obviously, randomized controlled clinical trialsare preferred. However, the vast majority of treatmentsavailable for children and adolescents have never beentested in any controlled or uncontrolled trial (Kazdin,1988). Understandably, we are encouraged if there is acrumb of data showing that someone changed somewhereafter exposure to treatment.

Finally, evidence from an outcome study that shows arelation between these processes hypothesized to becritical to therapeutic change and actual change would bevery persuasive. Assessment of processes might be re-flected in cognitions, family interaction or core conflictsand defenses. Therapeutic change would be shown tocovary with the extent to which these processes werealtered in treatment. This latter criterion is very de-manding indeed and perhaps is better conceived as a goaltoward which we strive rather than a point of departurefor identifying promising treatments.

No singie treatment among those available adequatelytraverses all of these criteria. Yet, a number of promisingtreatments have been identified for conduct disorder.Four treatment approaches with evidence on their behalfare illustrated next. In highlighting the approaches, thepurpose is not to convey that only four promisingtreatments exist. However, these four are clearly amongthe most well developed in relation to the criteriahighlighted here and the number of controlled clinicaltrials.^

Cognitive Probtem-sotving Skitls TrainingBaekground and underlying rationale. Cognitive pro-

cesses refer to a broad class of constructs that pertain tohow the individual perceives, codes and experiences theworld. Individuals who engage in conduct disorder

^ The rationale, empirical underpinnings, outcome researchand treatment procedures cannot be fully elaborated for each ofthe techniques. References will be made to reviews of theevidence and to treatment manuals that elaborate each of thetreatments.

164 A, E, KAZDIN

behaviors, particularly aggression, have been foutid toshow distortions and deficiencies in various cognitiveprocesses. These deficiencies are not merely reflections ofintellectual functioning. Although selected processes (re-call, infortnation processing) are related to intellectualfunctioning, their impact has been dehneated separatelyand shown to contribute to behavioral adjustment andsocial behavior.

A variety of cognitive processes have been studied,such as generating alternative solutions to interpersonalproblems (e.g. difl"erent ways of handhng social situ-ations), identifying the means to obtain particular ends(e.g. making friends) or consequences of one's actions(e.g. what could happen after a particular behavior);making attributions to others of the motivation of theiractions; perceiving how others feel; expectations oftheeffects of one's own actions and others (see Shirk, 1988;Spivack & Shure, 1982), Deficits and distortion amongthese processes relate to teacher ratings of disruptivebehavior, peer evaluations and direct assessment of overtbehavior (e.g, Lochman & Dodge, 1994; Rubin, Bream &Rose-Krasnor, 1991),

As an illustration, aggression is not merely triggered byenvironmental events, but rather through the way inwhich these events are perceived and processed. Theprocessing refers to the child's appraisals ofthe situation,anticipated reactions of others and self-statements inresponse to particular events. For example, attribution ofintent to others represents a salient cognitive dispositioncritically important to understanding aggressive be-havior. Aggressive youths tend to attribute hostile intentto others, especially in social situations where the cues ofactual intent are ambiguous (see Crick & Dodge, 1994).Understandably, when situations are initially perceivedas hostile, youths are more hkely to react aggressively.

Although many studies have shown that conduct-disordered youths experience various cognitive distor-tions and deficiencies, fundamental questions remain tobe resolved. Among these questions are the specificity ofcognitive deficits among diagnostic groups and youths ofdifferent ages, whether some of the processes are morecentral than others, and how these processes unfolddevelopmentally. Nevertheless, research on cognitiveprocesses among aggressive children has served as anheuristic base for conceptualizing treatment and fordeveloping specific treatment strategies.

Characteristics of treatment. Problem-solving skillstraining (PSST) consists of developing interpersonalcognitive problem-solving skills. Although many vari-ations of PSST have been applied to conduct problemchildren, several characteristics are usually shared. First,the emphasis is on how children approach situations, i.e.the thought processes in which the child engages to guideresponses to interpersonal situations. The children aretaught to engage in a step-by-step approach to solveinterpersonal problems. They make statements to them-selves that direct attention to certain aspects of theproblem or tasks that lead to effective solutions. Second,behaviors that are selected (solutions) to the interpersonalsituations are important as well. Prosocial behaviors arefostered (through modeling and direct reinforcement) aspart of the problem-solving process. Third, treatmentutilizes structured tasks involving games, academic ac-

tivities and stories. Over the course of treatment, thecognitive problem-solving skills are increasingly appliedto real-life situations. Fourth, therapists usually play anactive role in treatment. They model the cognitiveprocesses by making verbal self-statements, apply thesequence of statements to particular problems, providecues to prompt use ofthe skills and deliver feedback andpraise to develop correct use of the skills. Finally,treatment usually combines several different proceduresincluding modeling and practice, role-playing, and re-inforcement and mild punishment (loss of points ortokens). These are deployed in systematic ways to developincreasingly complex response repertoires of the child.

Overview of the evidence. Several outcome studieshave been completed with impulsive, aggressive andconduct-disordered children and adolescents (see Baer &Nietzel, 1991; Durlak, Furhman & Lampman, 1991 forreviews). Cognitively based treatments have significantlyreduced aggressive and antisocial behavior at home, atschool and in the community. At follow-up, these gainshave been evident up to one year later. Many earlystudies in the field (e.g. 1970s-80s) focused on impulsivechildren and nonpatient samples. Since that time, severalstudies have shown treatment effects with inpatient andoutpatient cases (see Kazdin, 1993; Kendall, 1991; Pepler& Rubin, 1991).

There is only sparse evidence that addresses the child,parent, family, contextual or treatment factors thatinfluence treatment outcome. Some evidence suggeststhat older children profit more from treatment than doyounger children, perhaps due to their cognitive de-velopment (Durlak et al,, 1991), However, the basis fordifferential responsiveness to treatment as a function ofage has not been well tested. Conduct-disordered childrenwho show comorbid diagnoses, academic delays anddysfunction and lower reading achievement, and whocome from families with high levels of impairment (parentpsychopathology, stress and family dysfunction) respondless well to treatment than youths with less dysfunction inthese domains (Kazdin, 1995a; Kazdin & Crowley, inpress). However, these child, parent and family charac-teristics may influence the effectiveness of several differenttreatments for conduct-disordered youths rather thanPSST in particular. Much further work is needed toevaluate factors that contribute to responsiveness totreatment.

Overall evaluation. There are features of PSST thatmake it an extremely promising approach. Perhaps mostimportantly, several controlled outcome studies withclinic samples have shown that cognitively based treat-ment leads to therapeutic change. Second, basic researchin developmental psychology continues to elaborate therelation of maladaptive cognitive processes among chil-dren and adolescents and conduct problems that serve asunderpinnings of treatment (Crick & Dodge, 1994; Shirk,1988). Third and on a more practical level, many versionsof treatment are available in manual form (e.g, Feindler& Ecton, 1986; Finch, Nelson & Ott, 1993; Shure, 1992).Consequently, the treatment can be evaluated in researchand explored further in clinical practice.

Fundamental questions about treatment remain. Tobegin, the role of cognitive processes in clinical dys-function and treatment warrant further evaluation. Evi-

TREATMENT OF CONDUCT DISORDER 165

dence is not entirely clear, showing that a specific patternof cognitive processes characterizes youths with conductproblems rather than adjustment problems more gen-erally. Also, although evidence has shown that cognitiveprocesses change with treatment, evidence has not es-tablished that change in these processes is correlated withimprovements in treatment outcome. This means that thebasis for therapeutic change has yet to be established.Also, characteristics of children and their families andparameters of treatment that may influence outcomehave not been carefully explored in relation to treatmentoutcome. Clearly, central questions about treatment andits efl"ects remain to be resolved. Even so, PSST is highlypromising because treatment eflects have been replicatedin several controlled studies with conduct-disorderedyouth.

Parent Management TrainingBackground and underlying rationale. Parent man-

agement training (PMT) refers to procedures in whichparents are trained to alter their child's behavior in thehome. The parents meet with a therapist or trainer whoteaches them to use specific procedures to alter inter-actions with their child, to promote prosocial behaviorand to decrease deviant behavior. Training is based onthe general view that conduct problem behavior isinadvertently developed and sustained in the home bymaladaptive parent-child interactions. There are multiplefacets of parent-child interaction that promote aggressiveand antisocial behavior. These patterns include directlyreinforcing deviant behavior, frequently and ineffectivelyusing commands and harsh punishment, and failing toattend to appropriate behavior (Patterson, 1982; Patter-son et al., 1992).

It would be misleading to imply that the parentgenerates and is solely responsible for the child-parentsequences of interactions. Influences are bidirectional,so that the child influences the parent as well (see Bell &Harper, 1977; Lytton, 1990). Indeed, in some cases thechildren engage in deviant behavior to help prompt theinteraction sequences. For example, when parents behaveinconsistently and unpredictably (e.g. not attending tothe child in the usual ways), the child may engage in somedeviant behavior (e.g. whining, throwing some object).The eflect is to cause the parent to respond in morepredictable ways (see Wahler & Dumas, 1986). Essen-tially, inconsistent and unpredictable parent behavior isan aversive condition for the child; the child's deviantbehavior is negatively reinforced by terminating thiscondition. However, the result is also to increase parentpunishment of the child.

Among the many interaction patterns, those involvingcoercion have received the greatest attention (Pattersonet al., 1992). Coercion refers to deviant behavior on thepart of one person (e.g. the child) that is rewarded byanother person (e.g. the parent). Aggressive children areinadvertently rewarded for their aggressive interactionsand their escalation of coercive behaviors, as part of thediscipline practices that sustain aggressive behavior. Thecritical role of parent-child discipline practices has beensupported by correlational research, relating specificdiscipline practices to child antisocial behavior, and by

experimental research, showing that directly alteringthese practices reduces antisocial child behavior (seeDishion, Patterson & Kavanagh, 1992).

The general purpose of PMT is to alter the pattern ofinterchanges between parent and child so that prosocial,rather than coercive, behavior is directly reinforced andsupported within the family. This requires developingseveral difl"erent parenting behaviors, such as establishingthe rules for the child to follow, providing positivereinforcement for appropriate behavior, delivering mildforms of punishment to suppress behavior, negotiatingcompromises and other procedures. These parentingbehaviors are systematically and progressively developedwithin the sessions in which the therapist shapes (developsthrough successive approximations) parenting skills. Theprograms that parents eventually implement in the homealso serve as the basis for the focus of the sessions inwhich the procedures are modified and refined.

Characteristics of treatment. Although many vari-ations of PMT exist, several common characteristics canbe identified. First, treatment is conducted primarily withthe parent(s), who implement several procedures in thehome. The parents meet with a therapist who teachesthem to use specific procedures to alter interactions withtheir child, to promote prosocial behavior and to decreasedeviant behavior. There is usually little direct interventionof the therapist with the child. With young children, thechild may be brought into the session to help train bothparent and child how to interact and especially to showthe parent precisely how to deliver prompts (antecedents)and consequences (reinforcement, time out from re-inforcement). Older youths may participate to negotiateand to develop behavior-change programs in the home.Second, parents are trained to identify, define and observeproblem behaviors in new ways. Careful specification ofthe problem is essential for the delivery of reinforcing orpunishing consequences and for evaluating if the programis achieving the desired goals. Third, the treatmentsessions cover social learning principles and the pro-cedures that follow from them including: positive re-inforcement (e.g. the use of social praise and tokens orpoints for prosocial behavior), mild punishment (e.g. useof time out from reinforcement, loss of privileges),negotiation, and contingency contracting. Fourth, thesessions provide opportunities for parents to see how thetechniques are implemented, to practise using the tech-niques, and to review the behavior-change programs inthe home. The immediate goal of the program is todevelop specific skills in the parents. As the parentsbecome more proficient, the program can address thechild's most severely problematic behaviors and en-compass other problem areas (e.g. school behavior). Overthe course of treatment, more complex repertoires aredeveloped, both in the parents and the child. Finally,child functioning at school is usually incorporated intothe program. Parent-managed reinforcement programsfor child deportment and performance at school, com-pletion of homework, activities in the playground and soon are often part of the behavior-change programs. Ifavailable, teachers can play an important role in moni-toring or providing consequences for behaviors at school.

Overview ofthe evidence. PMT is one ofthe most well-researched therapy techniques for the treatment of

166 A. E. KAZDIN

conduct-disordered youth. Scores of outcome studieshave been completed with youths varying in age anddegree of severity of dysfunction (e.g. oppositional,conduct disorder, delinquent youth) (see Kazdin, 1993;Miller & Prinz, 1990; Patterson, Dishion & Chamberlain,1993). Treatment effects have been evident in markedimprovements in child behavior on a wide range ofmeasures including parent and teacher reports of deviantbehavior, direct observation of behavior at home and atschool and institutional (e.g. school, police) records. Theeffects of treatment have also been shown to bringproblematic behaviors of treated children within nor-mative levels of their peers who are functioning ad-equately in the community. Follow-up assessment hasshown that the gains are often maintained 1-3 years aftertreatment. Longer follow-up assessment rarely takesplace, although one program reported maintenance ofgains 10-14 years later (Forehand & Long, 1988; Long,Forehand, Wierson & Morgan, 1994).

The impact of PMT is relatively broad. The effects oftreatment are evident for child behaviors that have notbeen focused on directly as part of training. Also, siblingsof children referred for treatment improve, even thoughthey are not directly focused on in treatment. This is animportant effect because siblings of conduct-disorderedyouths are at risk for severe antisocial behavior. Inaddition, maternal psychopathology, particularly depres-sion, has been shown to decrease systematically followingPMT (see Kazdin, 1985). These changes suggest thatPMT alters multiple aspects of dysfunctional families.

Several characteristics of the treatment contribute tooutcome. Duration of treatment appears to influenceoutcome. Brief and time-limited treatments (e.g. < 10hours) are less likely to show benefits with clinicalpopulations. More dramatic and durable effects havebeen achieved with protracted or time-unlimited pro-grams extending up to 50 or 60 hours of treatment (seeKazdin, 1985). Second, specific training components,such as providing parents with in-depth knowledge ofsocial learning principles and including time out fromreinforcement in the behavior-change program (in ad-dition to reinforcement) in the home, enhance treatmenteffects. Third, some evidence suggests that therapisttraining and skill are associated with the magnitude anddurability of therapeutic changes, although this has yet tobe carefully tested. Fourth, families characterized bymany risk factors associated with childhood dysfunction(e.g. socioeconomic disadvantage, marital discord, parentpsychopathology, poor social support) tend to showfewer gains in treatment than families without thesecharacteristics and to maintain the gains less well (e.g.Dadds & McHugh, 1992; Dumas & Wahler, 1983;Webster-Stratton, 1985). Some efforts to address parentand family dysfunction during PMT have led to improvedeffects of treatment outcome for the child in some studies(e.g. Dadds, Schwartz & Sanders, 1987; Griest et al.,1982) but not in others (Webster-Stratton, 1994). Muchmore work is needed on the matter, given the prominentrole of parent and family dysfunction among manyyouths referred for treatment.

One promising line of work has focused on implemen-tation of PMT in community, rather than chnic, settings.The net effect is to bring treatment to those persons least

likely to come to or remain in treatment. In one study, forexample, when PMT was delivered in small parent groupsin the community, the effectiveness surpassed what wasachieved with clinic-based PMT and was considerablymore cost effective (Cunningham, Bremner & Boyle,1995).

Conceptual development of processes underlyingparent-child interaction and conduct disorder continues(e.g. Patterson et al., 1992). Also, recent research onprocesses in treatment represents a related and importantadvance. A series of studies on therapist-parent in-teraction within PMT sessions has identified factors thatcontribute to parent resistance (e.g. parent saying, "Ican't," "I won't"). The significance of this work is inshowing that parent reactions in therapy relate to theirdiscipline practices at home, that changes in resistanceduring therapy predicts change in parent behavior andthat specific therapist ploys (e.g. reframing, confronting)can help overcome or contribute to resistance (Patterson& Chamberlain, 1994). This line of work advances ourunderstanding of PMT greatly by relating in-sessioninteractions of the therapist and parent to child func-tioning and treatment outcome.

O verall evaluation. Perhaps the most important pointto underscore is that no other technique for conductdisorder has probably been studied as often or as well incontrolled trials as has PMT. The outcome evidencemakes PMT one of the most promising treatments. Theevidence is bolstered by related lines of work. First, thestudy of family interaction processes that contribute toantisocial behavior in the home and evidence thatchanging these processes alters child behavior provide astrong empirical base for treatment. Second, the pro-cedures and practices that are used in PMT (e.g. variousforms of reinforcement and punishment practices) havebeen widely and effectively applied outside the context ofconduct disorder. For example, the procedures have beenapplied with parents of children with autism, languagedelays, developmental disabihties, medical disorders forwhich compliance with special treatment regimens isrequired and with parents who physically abuse or neglecttheir children (see Kazdin, 1994b). Third, a great deal isknown about the procedures and the parameters thatinfluence the reinforcement and punishment practicesthat form the core of PMT. Consequently, very concreterecommendations can be provided to change behaviorand to alter programs when behavior change has notoccurred.

A major advantage is the availability of treatmentmanuals and training materials for parents and pro-fessional therapists (e.g. Forehand & McMahon, 1981;Sanders & Dadds, 1993). Also noteworthy is the de-velopment of self-administered videotapes of treatment.In a programmatic series of studies with young conductproblem children (3-8 years), Webster-Stratton and hercolleagues have developed and evaluated videotapedmaterials to present PMT to parents; treatment can beself-administered in individual or group format supple-mented with discussion (e.g. Webster-Stratton, 1994;Webster-Stratton, Hollinsworth & Kolpacoff, 1989).Controlled studies have shown clinically significantchanges at post-treatment and follow-up assessmentswith variations of videotaped treatment. The potential

TREATMENT OF CONDUCT DISORDER 167

for extension of PMT with readily available and empiri-cally tested videotapes presents a unique feature in childtreatment.

Several limitations of PMT can be identified as well.First, some families may not respond to treatment. PMTmakes several demands on the parents, such as masteringeducational materials that convey major principles under-lying the program, systematically observing deviant childbehavior and implementing specific procedures at home,attending weekly sessions and responding to frequenttelephone contacts made by the therapist. For somefamilies, the demands may be too great to continue intreatment. Interestingly, within the approach severalprocedures (e.g. shaping parent behavior through re-inforcement) provide guidelines for developing parentcompliance and the desired response repertoire in relationto their children.

Second, perhaps the greatest hmitation or obstacle inusing PMT is that there are few training opportunities forprofessionals to learn the approach. Training programsin child psychiatry, clinical psychology, and social workare unlikely to provide exposure to the technique, muchless opportunities for formal training. PMT requiresmastery of social learning principles and multiple pro-cedures that derive from them (Cooper, Heron &Heward, 1987; Kazdin, 1994a). For example, the admin-istration of reinforcement by the parent in the home (toalter child behavior) and by the therapist in the session (tochange parent behavior) requires more than passingfamiliarity with the principle and the parametric vari-ations that dictate its effectiveness (e.g. need to administerreinforcement contingently, immediately, frequently, touse varied and high quality reinforcers; prompting,shaping). The requisite skills in administering these withinthe treatment sessions can be readily trained but they arenot trivial.

PMT has been applied primarily to parents of preado-lescents. Although treatment has been eflective withdelinquent adolescents (Bank, Marlowe, Reid, Patterson& Weinrott, 1991) and younger adolescents with conductproblems who have not yet been referred for treatment(Dishion & Andrews, 1995), some evidence suggests thattreatment is more efl"ective with preadolescent youths (seeDishion & Patterson, 1992). Parents of adolescents mayless readily change their discipline practices and also havehigher rates of dropping out of treatment. The im-portance and special role of peers in adolescence andgreater time that adolescents spend outside the homesuggest that the principles and procedures may need to beapplied in novel ways. At this point, few PMT programshave been developed specifically for adolescents, and soconclusions about the efi"ects for youths of different agesmust be tempered. On balance, PMT is one of the mostpromising treatment modalities. No other interventionfor conduct disorder has been investigated as thoroughlyas PMT.

Functional Family TherapyBackground and underlying rationale. Functional

family therapy (FFT) reflects an integrative approach totreatment that has relied on systems, behavioral andcognitive views of dysfunction (Alexander, Holtzworth-

Munroe & Jameson, 1994; Alexander & Parsons, 1982).Clinical problems are conceptualized from the standpointof the functions they serve in the family as a system, aswell as for individual family members. The assumption ismade that problem behavior evident in the child is theonly way some interpersonal functions (e.g. intimacy,distancing, support) can be met among family members.Maladaptive processes within the family are consideredto preclude a more direct means of fulfilling thesefunctions. The goal of treatment is to alter interactionand communication patterns in such a way as to fostermore adaptive functioning. Treatment is also based onlearning theory and focuses on specific stimuli andresponses that can be used to produce change. Social-learning concepts and procedures, such as identifyingspecific behaviors for change and reinforcing new adapt-ive ways of responding, and empirically evaluating andmonitoring change, are included in this perspective.Cognitive processes refer to the attributions, attitudes,assumptions, expectations and emotions of the family.Family members may begin treatment with attributionsthat focus on blaming others or themselves. New per-spectives may be needed to help serve as the basis fordeveloping new ways of behaving.

The underlying rationale emphasizes a family systemsapproach. Specific treatment strategies draw on findingsthat underlie PMT in relation to maladaptive andcoercive parent-child interactions, discussed previously.FFT views interaction patterns from a broader systemsview that also focuses on communication patterns andtheir meaning. As an illustration of salient constructs,research underlying FFT has found that families ofdelinquents show higher rates of defensiveness in theircommunications, both in parent-child and parent-parentinteractions, blaming and negative attributions, and alsolower rates of mutual support compared to families ofnondehnquents (see Alexander & Parsons, 1982). Im-proving these communication and support functions is agoal of treatment.

Characteristics of treatment. FFT requires that thefamily see the clinical problem from the relationalfunction it serves within the family. The therapist pointsout interdependencies and contingencies between familymembers in their day-to-day functioning and with specificreference to the problem that has served as the basis forseeking treatment. Once the family sees alternative waysof viewing the problem, the incentive for interacting moreconstructively is increased.

The main goals of treatment are to increase reciprocityand positive reinforcement among family members, toestablish clear communication, to help specify behaviorsthat family members desire from each other, to negotiateconstructively and to help identify solutions to inter-personal problems. In therapy, family members identifybehaviors they would hke others to perform. Responsesare incorporated into a reinforcement system in the hometo promote adaptive behavior in exchange for privileges.However, the primary focus is within the treatmentsessions, where family communication patterns arealtered directly. During the sessions, the therapist pro-vides social reinforcement (verbal and nonverbal praise)for communications that suggest solutions to problems,clarify problems or offer feedback.

168 A. E. KAZDIN

Overview of the evidence. Relatively few outcomestudies have evaluated FFT (see Alexander et al., 1994).However, the available studies have focused on difficultto treat populations (e.g. adjudicated delinquent adoles-cents, multiple offender delinquents) and have producedrelatively clear effects. In controlled studies, FFT has ledto greater change than other treatment techniques (e.g.client-centered family groups, psychodynamically orien-ted family therapy) and various control conditions (e.g.group discussion and expression of feeling, no-treatmentcontrol groups). Treatment outcome is reflected inimproved family communication and interactions andlower rates of referral to and contact of youth with thecourts. Moreover, gains have been evident in separatestudies up to 2| years after treatment.

Research has examined processes in therapy to identifyin-session behaviors of the therapist and how theseinfluence responsiveness among family members (Alex-ander, Barton, Schiavo & Parsons, 1976; Newberry,Alexander & Turner, 1991). For example, providingsupport and structure and reframing (recasting theattributions and bases of a problem) can influence familymember responsiveness and blaming of others. Therelations among such variables are complex insofar as theimpact of various type of statements (e.g. supportive) canvary as a function of gender of the therapist and familymember. Evidence of change in processes proposed to becritical to FFT (e.g. improved communication in treat-ment, more spontaneous discussion) supports the con-ceptual view of treatment.

Overall evaluation. Several noteworthy points can bemade about FFT. First, the outcome studies indicate thatFFT can alter conduct problems among delinquentyouth. Several studies have produced consistent effects.Second, the evaluation of processes that contribute tofamily member responsiveness within the sessions as wellas to outcome represents a line of work rarely seen amongtreatment techniques for children and adolescents. Someof this process work has extended to laboratory (analog)studies to examine more precisely how specific types oftherapist statements (e.g. reframing) can reduce blamingamong group members (e.g. Morris, Alexander & Turner,1991). Third, a treatment manual has been provided(Alexander & Parsons, 1982) to facilitate further evalu-ation and extension of treatment. Further work extendingFFT to children and to clinic populations would be ofinterest in addition to the current work with delinquentadolescents. Also, further work on child, parent andfamily characteristics that moderate outcome would be anext logical step in the existing research program.

Multisystemic TherapyBackground and underlying rationale. Multisystemic

therapy (MST) is a family-systems based approach totreatment (Henggeler & Borduin, 1990). Family ap-proaches maintain that clinical problems of the childemerge within the context of the family and focus ontreatment at that level. MST expands on that view byconsidering the family as one, albeit a very important,system. The child is embedded in a number of systemsincluding the family (immediate and extended familymembers), peers, schools, neighborhood and so on. For

example, within the context of the family, some tacitalliance between one parent and the child may contributeto disagreement and conflict over discipline in relation tothe child. Treatment may be required to address thealliance and sources of conflict in an effort to alter childbehavior. Also, child functioning at school may involvelimited and poor peer relations; treatment may addressthese areas as well. Finally, the systems approach entailsa focus on the individual's own behavior insofar as itaffects others. Individual treatment ofthe child or parentsmay be included in treatment.

Because multiple influences are entailed by the focus ofthe treatment, many different treatment techniques areused. Thus, MST can be viewed as a package ofinterventions that are deployed with children and theirfamilies. Treatment procedures are used on an "asneeded" basis directed toward addressing individual,family and system issues that may contribute to problembehavior. The conceptual view, focusing on multiplesystems and their impact on the individual, serves as abasis for selecting multiple and quite different treatmentprocedures.

Characteristics of treatment. Central to MST is afamily-based treatment approach. Several family therapytechniques (e.g. joining, reframing, enactment, paradoxand assigning specific tasks) are used to identify problems,increase communication, build cohesion, and alter howfamily members interact. The goals of treatment are tohelp the parents develop behaviors of the adolescent, toovercome marital difficulties that impede the parents'ability to function as parents, to eliminate negativeinteractions between parent and adolescent and to de-velop or build cohesion and emotional warmth amongfamily members.

MST draws on many other techniques as needed toaddress problems at the level of individual, family andextrafamily. As prominent examples, PSST, PMT, andmarital therapy are used in treatment to alter the responserepertoire of the adolescent, parent-child interactions athome and marital communication, respectively. In somecases, treatment consists of helping the parents address asignificant domain through practical advice and guidance(e.g. involving the adolescent in prosocial peer activitiesat school, restricting specific activities with a deviant peergroup). Although MST includes distinct techniques ofother approaches, it is not a mere amalgamation of them.The focus of treatment is on interrelated systems and howthey affect each other. Domains may be addressed intreatment (e.g. parent unemployment) because they raiseissues for one or more systems (e.g. parent stress, increasein alcohol consumption) and affect how the child isfunctioning (e.g. marital conflict, child discipline prac-tices).

Overview of the evidence. Several outcome studieshave evaluated MST, primarily with delinquent youthswith arrest and incarceration histories including violentcrime (e.g. manslaughter, aggravated assault with intentto kill). Thus, this is a group of extremely antisocial andaggressive youth. Results have shown MST to be superiorin reducing delinquency, emotional and behavioral prob-lems and in improving family functioning in comparisonto other procedures including "usual services" providedto such youths (e.g. probation, court-ordered activities

TREATMENT OF CONDUCT DISORDER 169

that are monitored such as school attendance), individualcounsehng and community-based eclectic treatment (e.g.Borduin et al., 1995; Henggeler et al., 1986; Henggeler,Melton, & Smith, 1992). Follow-up studies up to 2,4 and5 years later with separate samples have shown that MSTyouths have lower arrest rates than youths who receiveother services (see Henggeler, 1994).

Research has also shown that treatment affects criticalprocesses proposed to contribute to deviant behavior(Mann, Borduin, Henggeler, & Blaske, 1990). Specifi-cally, parents and teenage youths show a reduction incoalitions (e.g. less verbal activity, conflict and hostility)and increases in support, and the parents show increasesin verbal communication and decreases in conflict.Moreover, decreases in adolescent symptoms are posi-tively correlated with increases in supportiveness anddecreases in conflict between the mother and father. Thiswork provides an important link between theoreticalunderpinnings of treatment and outcome effects.

Overall evaluation. Several outcome studies are avail-able for MST and they are consistent in showing thattreatment leads to change in adolescents and that thechanges are sustained. A strength of the studies is thatmany of the youths who are treated are severely impaired(delinquent adolescents with a history of arrest). Anotherstrength is the conceptualization of conduct problems atmultiple levels, namely, as dysfunction in relation to theindividual, family and extrafamilial systems and thetransactions among these. In fact, youths with conductdisorder experience dysfunction at multiple levels in-cluding individual repertoires, family interactions andextrafamilial systems (e.g. peers, schools, employmentamong later adolescents). MST begins with the view thatmay different domains are likely to be relevant; they needto be evaluated and then addressed as needed in treat-ment.

A challenge of the approach is deciding what treat-ments to use in a given case, among the many inter-ventions encompassed by MST. Guidehnes are availableto direct the therapist, although they are somewhatgeneral (e.g. focus on developing positive sequences ofbehaviors between systems such as parent and adolescent,evaluate the interventions during treatment so thatchanges can be made; see Henggeler, 1994). Providinginterventions as needed is very difficult without a con-sistent way to assess what is needed, given inherent limitsof decision making and perception, even among trainedprofessionals. Related to this, the administration of MSTis demanding in light of the need to provide severaldifferent interventions in a high-quality fashion. In-dividual treatments (e.g. PSST, PMT) alone are difficultto provide; multiple combinations invite problems relatedto providing treatments of high quahty, strength andintegrity. Yet there have been replications of MST beyondthe original research program, indicating that treatmentcan be extended across therapists and settings (Henggeler,Schoenwald & Pickrel, 1995).

On balance, MST is quite promising given the qualityof evidence and consistency in the effects that have beenproduced. The promise stems from a conceptual ap-proach that examines multiple domains (systems) andtheir contribution to dysfunction, evidence on processesin therapy and their relation to outcome and the outcome

studies themselves. The outcome studies have extended toyouths with different types of problems (e.g. sexualoffenses, drug use) and to parents who engage in physicalabuse or neglect (e.g. Borduin, Henggeler, Blaske &Stein, 1990; Brunk, Henggeler & Whelan, 1987). Thus,the model of providing treatment may have broadapplicability across problem domains among seriouslydisturbed children. In passing, it may be worth notingthat other literatures are relevant to MST. Some of thetechniques included in treatment are variations of PSSTand PMT, already discussed, and hence have evidence ontheir own behalf as effective interventions.

Limitations of Promising TreatmentsEach of the treatments just discussed has randomized,

controlled trials on its behalf, includes replications oftreatment effects in multiple studies, focuses on youthswhose aggressive and antisocial behavior have led toimpairment and referral to social services (e.g. clinics,hospitals, courts) and has assessed outcome over thecourse of follow-up, at least up to a year, but often longer.Even though these treatments have made remarkablegains, they also bear limitations worth highlighting.

Magnitude of therapeutic change. Promising treat-ments have achieved change, but is the change enough tomake a difference in the lives of the youths who aretreated? Clinical significance refers to the practical valueor importance of the effect of an intervention, that is,whether it makes any "real" difference to the patients orto others with whom they interact (see Kazdin, 1992).Clinical significance is important because it is quitepossible for treatment effects to be statistically significant,but not to have impact on most or any of the cases in away that improves their functioning or adjustment indaily life.

There are several ways to evaluate clinical significance.As an example, one way is to consider the extent to whichyouths function at normative levels at the end oftreatment (i.e. compared to same age and sex peers whoare functioning well). This is particularly useful as acriterion in relation to children and adolescents becausebase rates of emotional and behavioral problems can varygreatly as a function of age. Promising treatmentsoccasionally have shown that treatment returns indi-viduals to normative levels in relation to behavioralproblems and prosocial functioning at home and atschool (see Kazdin, 1995b). Yet, the majority of studies,whether of promising or less well-evaluated treatments,have not examined whether youths have changed in waysthat place them within normative range of functioning orhave made gains that would reflect clinically significantchanges (Kazdin, Bass, Ayers & Rodgers, 1990a).

Although the goal of treatment is to effect clinicallysignificant change, other less dramatic goals are nottrivial. For many conduct-disordered youths, symptomsmay escalate, comorbid diagnoses (e.g. substance abuse,depression) may emerge and family dysfunction mayincrease. Also, such youths are at risk for teen marriage,dropping out of school and running away. If treatmentwere to achieve stability in symptoms and family life andprevent or dehmit future dysfunction, that would be asignificant achievement. The reason evaluation is so

170 A. E. KAZDIN

critical to the therapeutic enterprise is to identify whethertreatment makes a difference because "making a dif-ference" can have many meanings that are important inthe treatment of conduct disorder.

Maintenance of change. Promising treatments haveincluded follow-up assessment, usually up to a year aftertreatment. Yet, conduct disorder has a poor long-termprognosis, so it is especially important to identify whethertreatment has enduring effects. Also, in evaluating therelative merit of different treatments, follow-up data playa critical role. When two (or more) treatments arecompared, the treatment that is more (or most) effectiveimmediately after treatment is not always the one thatproves to be the most effective treatment in the long run(Kazdin, 1988). Consequently, the conclusions abouttreatment may be very different depending on the timingof outcome assessment. Apart from conclusions abouttreatment, follow-up may provide important infonnationthat permits differentiation among youths. Over time,youths who maintain the benefits of treatment may differin important ways from those who do not. Understandingwho responds and who responds more or less well to aparticular treatment can be very helpful in understanding,treating and preventing conduct disorder.

The study of long-term effects of treatment is difficultin general, but the usual problems are exacerbated byfocusing on conduct disorder. Among clinic samples,families of conduct-disordered youths have high rates ofdropping out during treatment and during the follow-upassessment period, due in part to the many parent andfamily factors (e.g. sodoeconomic disadvantage, stress)often associated with the problem (Kazdin, 1996b). Asthe sample size decreases over time, conclusions about theimpact of treatment become increasingly difficult to draw.Nevertheless, evaluation of the long-term effects oftreatment remains a high priority for research.

Limited assessment of outcome domains. In the ma-jority of child therapy studies, child symptoms are theexclusive focus of outcome assessment (Kazdin et al.,1990a). Other domains such as prosocial behavior andacademic functioning are neglected, even though theyrelate to concurrent and long-term adjustment (e.g. Asher& Coie, 1990). Perhaps the greatest single deficit in theevaluation of treatment is absence of attention to im-pairment. Impairment refiects the extent to which theindividual's functioning in everyday life is impeded.Impairment can be distinguished from symptoms insofaras individuals with similar levels of symptoms (e.g.scores), diagnoses and patterns of comorbidity are likelyto be distinguishable based on their ability to functionadaptively. School and academic functioning, peer rela-tions, participation in activities and health are some ofthe areas included in impairment. In the context oftreatment, an intervention may significantly reduce symp-toms. Yet, is there any change or reduction in im-pairment? The impact of treatment on impairment isarguably as important as the impact on the conductdisorder symptoms.

Beyond child functioning, parent and family function-ing may also be relevant. Parents and family members ofconduct-disordered youths often experience dysfunction(e.g. psychiatric impairment, marital confiict). Also, theproblem behaviors ofthe child are often part of complex.

dynamic and reciprocal infiuences that affect all relationsin the home. Consequently, parent and family functioningand the quality of life for family members are relevantoutcomes and may be appropriate goals for treatment.

In general, there are many outcomes that are of interestin evaluating treatment. From existing research wealready know that the conclusions reached about a giventreatment can vary depending on the outcome criterion.Within a given study, one set of measures (e.g. childfunctioning) may show no differences between twotreatments but another measure (e.g. family functioning)may show that one treatment is clearly better than theother (e.g. Kazdin, Bass, Siegel & Thomas, 1989; Kazdin,Siegel & Bass, 1992; Szapocznik et al., 1989). Thus, inexamining different outcomes of interest, we must beprepared for different conclusions that these outcomesmay yield.

General CommentsIn light of these comments, clearly even the most

promising treatments have several limitations. Yet it iscritical to place these in perspective. The most commonlyused treatments in clinical practice consist of "tra-ditional " approaches including psychodynamic, relation-ship, play and family therapies (other than those men-tioned earlier) (Kazdin, Siegel & Bass, 1990b). Thesetreatments have rarely been tested in controlled outcomestudies showing that they achieve therapeutic change inreferred (or nonreferred) samples of youth with conductproblems. Many forms of behiavior therapy have a ratherextensive literature showing that various techniques (e.g.reinforcement programs, social skills training) can alteraggressive and other antisocial behaviors (Kazdin, 1985;McMahon & Wells, 1989). Yet the focus has tended to beon isolated behaviors, rather than a constellation ofsymptoms. Also, durable changes among clinical sampleshave rarely been shown.

Pharmacotherapy represents a line of work of someinterest. For one reason, stimulant medication (e.g.methylphenidate), frequently used with children diag-nosed with ADHD, has some impact on aggressive andother antisocial behaviors (see Hinshaw, 1994). This isinteresting in part because such children often have acomorbid diagnosis of Conduct Disorder. Still no strongevidence exists that stimulant medication can alter theconstellation of symptoms (e.g. fighting, steahng) associ-ated with conduct disorder. A review of various medi-cations for aggression in children and adolescents hasraised possible leads, but the bulk of research consists ofuncontrolled studies (see Campbell & Cueva, 1995;Stewart, Myers, Burket & Lyles, 1990). Controlled studies(e.g. random assignment, placebo-controls) have shownantiaggressive effects with some medications (e.g. lith-ium; Campbell et al., 1995) but not others (e.g. carba-mazepine; Cueva et al., 1996). Reliable psychopharmaco-logical treatments for aggression, leaving aside theconstellation of conduct disorder (e.g. firesetting, steal-ing, and so on), remain to be developed.

There is a genre of interventions that are worthmentioning but are even less well evaluated than many ofthe psychotherapies and pharmacotherapies. Occa-sionally, interventions are advocated and implemented.

TREATMENT OE CONDUCT DISORDER 171

such as sending conduct-disordered youths to a camp inthe country where they learn how to "rough it," or howto take care of horses or to experience mihtary (e.g. basictraining) regimens. The conceptual bases of such treat-ments, research identifying processes involved in theonset of conduct disorders and related criteria (notedearlier in the paper) are rarely even approximated withthis genre of interventions. Typically, such programs arenot evaluated empirically. On the one hand, developingtreatments that emerge outside of the mainstream of themental health professions is to be encouraged preciselybecause traditional treatments have not resolved theproblem. On the other hand, this genre of interventiontends to avoid evaluation. Evaluation is the key becausewell-intentioned and costly interventions can have littleor no effect on the youths they treat (Weisz, Walter,Weiss, Fernandez & Mikow, 1990) and may actuallyincrease antisocial behavior (e.g. see Lundman, 1984).

Salient Clinical Issues in Treatment

There are a number of issues that emerge in treatmentof conduct-disordered youths and decision-making aboutwhat interventions to provide to whom. These issuesrefiect obstacles in delivering treatment, lacunae in ourknowledge base and limitations in the models of pro-viding care.

Retaining Cases in TreatmentDropping out from treatment is a significant problem

in the treatment of children and adolescents.'^ Amongfamilies that begin treatment, 40-60 % terminate prema-turely (Armbruster & Kazdin, 1994; Wierzbicki & Peka-rik, 1993). Youths with aggressive and antisocial behaviorare particularly likely to drop out early (e.g. Capaldi &Patterson, 1987; Kaminer, Tarter, Bukstein & Kabene,1992).

Many ofthe parent and family factors often associatedwith conduct disorder are likely to place families at riskfor terminating treatment prematurely. These include:socioeconomic disadvantage, facets of the family con-stellation (younger mothers, single-parent families), highparent stress, adverse child-rearing practices (e.g. harshpunishment, poor monitoring and supervision of thechild) and parent history of antisocial behavior (e.g.Kazdin, 1990; Kazdin, Mazurick & Bass, 1993;McMahon, Forehand, Griest & Wells, 1981). Childcharacteristics that predict early termination from treat-ment include comorbidity (multiple diagnoses and symp-toms across a range of disorders), severity of delinquentand antisocial behavior and poor academic functioning.The accumulation of these factors places families at

^ Dropping out of treatment usually refers to prematurelyterminating from ongoing therapy at a point where the patientceases to come for treatment and when the therapist believesthat this decision is ill-advised. In research, early terminationusually refers to dropping out within the first few sessions oftreatment, although the patient may leave the system at manydifferent points (e.g. after being referred to the clinic, contactingthe clinic by phone, scheduling an initial appointment, attendingthat appointment, beginning intake assessment).

increased risk for dropping out of treatment within thefirst few sessions. Interestingly, many ofthe child, parentand family factors that predict premature terminationfrom treatment are the same factors that portend a poorresponse to treatment and poor long-term prognosis(Dadds & McHugh, 1992; Dumas & Wahler, 1983;Kazdin, 1995a; Webster-Stratton, 1985).

The cases who terminate early are those who evince thegreatest impairment in parent, family and child charac-teristics. In clinical work, the usual impression is thatindividuals who drop out of treatment are much worse offthan those who have remained in treatment. Our ownwork suggests that this is true, but due primarily to thefact that those who drop out are more severely impairedto begin with (Kazdin, Mazurick & Siegel, 1994). Evenso, evidence points to benefits of remaining in treatment.Those cases who remain in treatment but are equallyimpaired as those who have dropped out tend to farebetter. Consequently, it is important to retain cases intreatment.

Even though treatment is designed to help families,several aspects of coming to treatment increase stress anddemands on the family. Many of the burdens areassociated with coming to the sessions and includeprocuring transportation, cajoling the identified patient(child) to agree to come to the session that day, arrangingbabysitting for other children and so on. In fact, parentswill often cancel a session or not show up because of thedifficulties of bringing the child and the child's sibhngs tothe clinic. Financial costs associated with coming totreatment (e.g. babysitting, transportation, costs of treat-ment) also may be a significant burden in light of thedisproportionate distribution of poverty among familiesof youths referred to treatment for conduct disorder.

There are a few leads for retaining cases better intreatment, although the empirical evidence in relation tochild and adolescent treatment is sparse. Providingspecial sessions for the parents to address sources of stressand concern (e.g. job stress, personal worries, familydisputes), when added to treatment of the child, reducesattrition (Prinz & Miller, 1994). Also, providing childrenwith a special preparatory interview to convey why peoplego to therapy reduces the rate of dropping out (Holmes &Urie, 1975). Developing an alliance with all immediatefamily members (e.g. by extensive phone contacts) earlyin treatment, conveying the benefits that can accrue toeach member as the child improves and making an effortto engage the family members in treatment as obstaclesemerge have reduced attrition and improved treatmentoutcome (Santisteban et al., 1996; Szapocznik et al.,1988).

Clinically, adding interventions just to retain cases intreatment can place a burden on the therapy. Yet cases athigh risk for dropping out can be identified, based onfactors mentioned previously. In these cases, it may befeasible for the chnician to attack both fronts, namely,improvement of the conduct-disordered child and re-duction of parental stress and the burden of treatment. Itis unlikely that improvements in child functioning alonewill help retain cases in treatment, given what we knowabout the factors that predict treatment termination. Infact, in our own clinical work, early improvement intreatment seems to increase the likelihood of attrition.

172 A, E, KAZDIN

Some parents perceive early changes as sufficient, eventhough many areas may require further attention.

What Treatments Do Not WorkWith a few hundred or so treatments available for

children, it would be quite helpful to know which amongthese do not work or do not work very well. Addressingthe matter directly is not possible in light of the fact,noted previously, that the vast majority of treatmentapproaches have not been evaluated empirically. Thus,there is no accumulated body of evidence in whichtreatments have consistently emerged as weak or inef-fective. Moreover, the nature of the dominant scientificresearch paradigm (inability to prove the null hypothesis)precludes firm demonstration of no effects of treatment.Most of the treatments currently used in clinical work(Kazdin et al., 1990b), including psychodynamic therapy,relationship-based treatment and play therapy, have notbeen evaluated empirically (Kazdin et al., 1990a). Occa-sionally, variations of these treatments have been used ascomparative conditions and have been shown to be lesseffective than one of the promising treatments notedpreviously (e.g. Borduin et al., 1995; Kazdin, Esveldt-Dawson, French & Unis, 1987a, b). From this limitedresearch, it is premature to conclude that these lattertreatments are ineffective. Yet at best their benefits havestill to be demonstrated and more promising treatmentswith firmer empirical bases are currently the treatmentsof choice.

The absence of empirical evidence is only one criterion,albeit an obviously important one. In advance of, andeventually along with, the evidence, scrutiny of theconceptual underpinnings of treatment and the treatmentfocus is important in relation to what we know aboutconduct disorder. We know, for example, that conduct-disordered youths usually show problems in multipledomains, including overt behavior, social relations (e.g.peers, teachers, family members) and academic per-formance. For a treatment to be effective, it is hkely thatseveral domains have to be addressed explicitly within thesessions or a conceptual model (with supporting evidence)is needed to convey why a narrow or delimited focus (e.g.on psychic conflicts or a small set of overt behaviors) islikely to have broad effects on domains not explicitlyaddressed in treatment. Although one cannot say forcertain what techniques will not work, it is much safer tosay that treatments that neglect multiple domains arelikely to have limited effects.

Second, some evidence has emerged that is useful forselecting what treatments to avoid or to use with greatcaution. Often conduct-disordered youths are treated ingroup therapy, yet placing youths together could impedeimprovement. For example, Feldman, Caphnger andWodarski (1983) randomly assigned youths (ages 8-17)to variations of group therapy. In one type of group, allmembers were referred for conduct disorder; in anothertype of group, conduct-disordered youths were placedwith nonantisocial youths (without clinical problems).Those placed in a group of their deviant peers did notimprove; those placed with nondeviant peers did im-prove. Interpretation of this is based on the likelihood

that peer bonding to others can improve one's behavior,if those peers engage in more normative behavior;bonding to a deviant group can sustain deviant behavior.

Similarly, Dishion and Andrews (1995) evaluatedseveral interventions for nonreferred youths (ages 10-14)with conduct problems. One of the treatment conditionsincluded youths meeting in a group with a focus on self-regulation, monitoring and developing behavior-changeprograms. This condition, whether alone or in com-bination with parent training, was associated with in-creases in behavioral problems and substance use (ciga-rette smoking). Again, it appeared that placing conduct-problem teens in a group situation can exacerbate theirproblems. Other research has shown that individuals maybecome worse (e.g, increase in arrest rates) througliassociation with deviant peers as part of treatment(O'Donnell, 1992).

Treatments for conduct-disordered youths in settingssuch as hospitals, schools and correctional facilities areoften conducted in a group therapy format in whichseveral conduct-problem youths are together to talkabout or work on their problems or go to the country forsome fresh air experience to get better. There may beconditions under which this arrangement is beneficial.However, current research suggests that placing severalsuch youths together can impede therapeutic change andhave deleterious effects.

Who Responds Well to TreatmentWe have known for many years that the critical

question of psychotherapy is not what technique iseffective, but rather what technique works for whom,under what conditions, as administered by what type oftherapists, and so on (Kiesler, 1971). The adult psycho-therapy literature has f"ocused on a range of questions toidentify factors (e.g. patient, therapist, treatment process)that contribute to outcome. The child and adolescenttherapy research has been devoted almost exclusively toquestions about treatment technique, with scant attentionto the role of child, parent, family and therapist factorsthat may moderate outcome (Kazdin et al,, 1990a).

In the case of conduct disorder, a few studies havelooked at who responds to treatment, mostly in thecontext of parent management training and problem-solving skills training. Current evidence suggests thatrisk factors for onset of conduct disorder and poorlong-term prognosis also predict response to treatment(Dumas & Wahler, 1983; Kazdin, 1995a, Kazdin &Crowley, in press; Webster-Stratton, 1985). Multiplechild, parent, family and contextual factors, includingearly onset and more severe child antisocial behavior,comorbid diagnoses, child academic impairment, socio-economic disadvantage, single-parent families, parentalstress (perceived) and life events, and parent history ofantisocial behavior in childhood are likely to infiuenceresponsiveness to treatment. These factors accumulateand increase risk for poor outcome in treatment. Our ownwork has shown that even those youths with multiple riskfactors still improve with treatment, but the changes arenot as great as those achieved for cases with fewer riskfactors. The characteristics that have been studied in

TREATMENT OF CONDUCT DISORDER 173

relation to treatment outcome (e.g. comorbidity) havenot been examined across different treatments. Conse-quently, we do not know whether these factors affectresponsiveness to any treatment or to particular forms oftreatment.

In current subtyping of conduct-disordered youths,early (childhood) and later (adolescent) onset conductdisorder are distinguished (Hinshaw et al., 1993; Moffitt,1993), Early-onset conduct-disordered youths are charac-terized by aggressive behavior, neuropsychological dys-function (in "executive" functions), a much higher ratioof boys to girls and a poor long-term prognosis. Later-onset youths (onset at about age 15) are characterizedmore by delinquent activity (theft, vandalism), a moreeven distribution of boys and girls and a more favorableprognosis. The subtype and associated characteristics areby no means firmly established, but refiect currentconceptual and empirical work in the area (e.g. Moffitt,1993; Patterson, DeBaryshe & Ramsey, 1989). We canexpect from this that youths with an early onset are morelikely to be recalcitrant to treatment. At present, and inthe absence of very much treatment research on thematter, a useful guideline to predict responsiveness totreatment is to consider loading of the child, parent andfamily on risk factors that portend a poor long-termprognosis (see Kazdin, 1995b; Robins, 1991).

In clinical work, there is frequent discussion about theimportance of individualizing treatment to the needs ofthe child and family. At this point, the research is of littlehelp in addressing the level of specificity in craftingtreatment regiments to the individual. A possible ex-ception is one of the treatments mentioned previously(multisystemic therapy), in which several different treat-ments, some with firm evidence on their behalf, areintegrated as a treatment package. At present, perhapsthe best strategy is to select the treatment that appears tobe promising based on the evidence and applying that asthe initial treatment of choice. Attempting to makedecisions about what can be applied effectively amongpromising or unpromising techniques is difficult to do inan informed way in light of the current knowledge baseand could very well lead to less effective clinical care forthe individual child and family.

Addressing ComorbidityAn issue that has received attention in discussions of

clinical dysfunction and treatment is the issue of comor-bidity. As noted previously, conduct disorder is oftencomorbid with other diagnoses, most notably ADHDand ODD, but others as well. It is hkely that comorbidityis the rule rather than the exception among cases referredfor treatment. In our own clinic, for example, approxi-mately 70 % of the cases meet DSM criteria for two ormore disorders (Kazdin, 1996b).

Comorbidity has been conceived of rather narrowly,namely, the presence of two or more disorders. In relationto treatment research and practice, there may be value inextending the notion more broadly. It is likely thatchildren have many symptoms from many differentdisorders, even though they might not meet the criteriafor each ofthe disorders. Indeed, in our research we have

found the total number of symptoms across the range ofdisorders to be a more sensitive predictor of treatmentoutcome than merely counting the number of diagnoses(Kazdin & Crowley, in press). Although the number ofdisorders may be important, impairment across the fullrange of symptoms is noteworthy as well.

It may be useful to expand the notion of comorbiditywell beyond symptoms and diagnoses. A central issue fortreating conduct-disordered youth is the domains ofimpairment they experience. These domains can includeother disorders (e.g. depression, substance abuse), learn-ing difficulties (specific reading disorders, language de-lays, learning disability), dysfunctional peer relations(rejection, absence of prosocial friends) and perhapsdeficits in prosocial activities (participation in school,athletic and extracurricular events). Problems or dysfunc-tions in each of these domains, apart from conduct-disorder symptoms themselves, can influence the effectsof treatment and long-term prognosis.

At present, research has not provided guidelines forhow to address comorbid conditions. Indeed, much ofthetreatment research has eschewed diagnosis, so the numberor proportions of youth who meet criteria for anydisorder is usually unclear (Kazdin et al., 1990a). We cansay very little at this point about whether comorbidconditions invariably influence outcome, whether theinfluence and direction of that infiuence vary by thespecific comorbid condition, or how to alter treatment inhght of these conditions. This area of work represents amajor deficiency in the knowledge base among even themost promising treatments for conduct disorder.

Combining TreatmentsThere is keen interest, both in clinical work and in

research, in using combinations of treatment, i.e. multiplepsychosocial and/or pharmacological interventions (seeKazdin, 1996a). In the case of conduct disorder, impetusstems from the scope of impairment evident in children(e.g. comorbidity, academic dysfunction) and families(e.g. stress, conflict) as well as limited effects of mosttreatments. The benefits of combined treatments can beidentified in selected areas. For example, in the treatmentof adult schizophrenia, combinations of treatment (e.g.medication and family counseling/therapy) surpasses theeffects ofthe constituent components alone (e.g. Falloon,1988).

In the case of child and adolescent therapy, combinedtreatments have not been well studied. I have arguedelsewhere that there are many reasons to expect combinedtreatments not to surpass the effects of any promisingsingle treatment (Kazdin, 1996a). Among the reasons, weknow very little about the parameters of a given treatmentthat influence its effectiveness and the cases to whom thetreatment is most suitably apphed. Combining techniquesof which we know relatively little, particularly in time-limited treatment, is not a firm base to build moreeffective treatments. Also, there are many obstacles incombining treatment that materially affect their likelyoutcome, such as decision rules regarding what treat-ments to combine, how to combine them (e.g. when, inwhat order), how to evaluate their impact and others.

174 A, E, KAZDIN

An important assumption for combined treatments isthat individual treatments are weak and, if combined,they would produce additive or synergistic effects. This isa reasonable, even if poorly tested, assumption. Analternative assumption is that the way in which treatmentis usually administered, whether a single or a combinedtreatment, inherently limits the likehhood of positiveoutcome effects, a point discussed further later. As ageneral point, combining treatments itself is not likely tobe an answer to developing effective treatment withoutmore thought and evidence about the nature of thesecombinations.

Some of the promising treatments reviewed previously(MST, FFT) are combined treatments. For example,multisystemic therapy provides many different treatmentsfor antisocial youths. Two points are worth noting. First,the constituent treatments that form a major part oftreatment are those that have evidence on their behalf(e,g, PSST, PMT), so that not any combination is used.Second, we do not yet know that multisystemic therapy,as a combined treatment package, is more effective thanthe. most effective constituent component administeredfor the same duration. The comparisons of multisystemictherapy have mostly included ordinary individual psycho-therapy and counseling, important comparison groups tobe sure. Although treatment has surpassed traditionaltherapy practices, this is not the same as showing thatcombinations of treatment per se are necessary to achievetherapeutic changes.

Combined treatments may be very useful and shouldbe pursued. At the same time, a rash move to combinetreatments is unwarranted. The effects of combinedtreatment obviously depend very much on the individualtreatments that are included in the combination. Forexample, mentioned already was a study in which parenttraining and a teen-focused group were evaluated aloneand in combination (Dishion & Andrews, 1995). Con-ditions that received the teen-group component, whetheralone or in combination with parent training, becameworse. Obviously, one cannot assume that combinedtreatments will automatically be neutral or better thantheir constituent treatments. There is another more subtleand perhaps worrisome facet of combined treatments, Adanger in promoting treatment combinations is to con-tinue to use techniques with little evidence on their behalfas an ingredient in a larger set of techniques. Old wine innew bottles is not bad if the original wine has merit.However, without knowing if there is merit, the tendencyto view the wine as new and improved would beunfortunate. With promising treatments available, wehave a comparative base to evaluate novel treatments,treatment combinations, and unevaluated treatments incurrent use. If a promising treatment is not used inclinical work, we would want evidence that it has clearlyfailed, that other promising treatments for whateverreason cannot be used and that the treatment that is to beapplied has a reasonable basis for addressing the scope ofdysfunctions.

Models of Delivering TreatmentThe model of treatment delivery in current research is

to provide a relatively brief and time-hmited intervention.

For several clinical dysfunctions or for a number ofchildren with a particular dysfunction such as conductdisorder, the course of maladjustment may be long-term.In such cases, the notion of providing a brief, time-hmitedtreatment may very much limit outcome effects. Even if agre^t combination of various psychotherapies were con-structed, administration in the time-limited fashion mighthave the usual, checkered yield. More extended andenduring treatment in some form may be needed toachieve clinically important effects with the greatestnumber of youths. Two ways of delivering extendedtreatment illustrate the point.

The first variation might be referred to as a continued-care model. The model of treatment delivery that may beneeded can be likened to the model used in the treatmentof diabetes mellitus. With diabetes, ongoing treatment(insulin) is needed to ensure that the benefits of treatmentare sustained. The benefits of treatment would end withdiscontinuation of treatment. Analogously, in the contextof conduct disorder, a variation of ongoing treatmentmay be needed. Perhaps after the child is referred,treatment is provided to address the current crises and tohave impact on functioning at home, at school and in thecommunity. After improvement is achieved, treatment ismodified rather than terminated. At that point, the childcould enter into maintenance therapy, i.e. continuedtreatment perhaps in varying schedules ("doses"). Treat-ment would continue but perhaps on a more intermittentbasis. Continued treatment in this fashion has beeneffective as a model for treating recurrent depression inadults (see Kupfer et al,, 1992).

The second variation might be referred to as a dental-care model to convey a different way of extendingtreatment. After initial treatment and demonstratedimprovement in functioning in everyday hfe, treatment issuspended. At this point, the child's functioning begins tobe monitored regularly (e.g, every 3 months) and sys-tematically (with standardized measures). Treatmentcould be provided pro re nata (PRN) based on theassessment data or emergent issues raised by the family,teachers or others. The approach might be likened to themore famihar model of dental care in the United States inwhich "check-ups" are recommended every 6 months;an intervention is provided if and as needed, based onthese periodic checks.

Obviously, the use of ongoing treatment is not advo-cated in cases where there is evidence that short-termtreatment is effective. A difficulty with most of theresearch on treatment of conduct disorder, whetherpromising, poorly investigated or combined treatments,is that the conventional treatment model of brief, time-limited therapy has been adopted. Without consideringalternative models of delivery, current treatments may bequite hmited in the effects they can produce. Althoughmore effective treatments are sorely needed, the way ofdelivering currently available treatments ought to bereconsidered.

ConclusionsMany different types of treatment have been applied to

conduct-disordered youths. Unfortunately, little out-come evidence exists for most of the techniques. Four

TREATMENT OF CONDUCT DISORDER 175

treatments with the most promising evidence to date werehighlighted: problem-solving skills training, parent man-agement training, functional family therapy, and multi-systemic therapy. Cognitive problem-solving skills train-ing focuses on cognitive processes that underlie socialbehavior. Parent management training is directed ataltering parent-child interactions in the home, particu-larly those interactions related to child-rearing practicesand coercive interchanges. Functional family therapyutilizes principles of systems theory and behavior modifi-cation as the basis for altering interactions, communi-cation and problem solving among family members.Multisystemic therapy focus on the individual, familyand extrafamilial systems and their interrelations as away to reduce symptoms and to promote prosocialbehavior. Evidence on behalf of these interventions wasreviewed; each has multiple controlled studies on itsbehalf and some ofthe techniques (e.g. PMT) have beenextraordinarily well evaluated.

Significant issues remain to be addressed to accelerateadvances in the area of treatment. The magnitude ofchange and durability of treatment effects raise multipleissues about how to evaluate treatment and the con-clusions reached about any particular intervention. Wecannot yet say that one intervention can ameliorateconduct disorder and overcome the poor long-termprognosis. On the other hand, much can be said. Much ofwhat is practised in clinical settings is based on psycho-dynamically oriented treatment, general relationshipcounseling, family therapy and group therapy (withantisocial youths as members). These and other pro-cedures, alone and in various combinations in which theyare often used, have not been evaluated carefully incontrolled trials. Of course, absence of evidence is nottantamount to ineffectiveness. At the same time, prom-ising treatments have advanced considerably and a veryspecial argument might be needed to administer treat-ments that have neither basic research on their conceptualunderpinnings in relation to conduct disorder nor out-come evidence from controlled clinical trials on theirbehalf. Promising treatments, at best, leave importantquestions unanswered. Further development of treat-ments is clearly needed. Apart from treatment studies,further progress in understanding the nature of conductdisorder is likely to have very important implications forimproving treatment outcome. Improved triage ofpatients to treatments that are likely to work will requireunderstanding of characteristics of children, parents andfamilies that will make them more or less amenable tocurrent treatments.

Acknowledgements—Completion of this paper was supportedby a Research Scientist Award (MH00353) and a grant(MH35408) from the National Institute of Mental Health.Support for this work is gratefully acknowledged.

ReferencesAlexander, J. F., Barton, C , Schiavo, R. S. & Parsons, B. V.

(1976). Systems-behavioral intervention with families ofdelinquents: Therapist characteristics, family behavior, andoutcome. Journal of Consulting and Clinical Psychology, 44,656-664.

Alexander, J. F., Holtzworth-Munroe, A. & Jameson, P. B.(1994). The process and outcome of marital and familytherapy research: Review and evaluation. In A. E. Bergin &S. L. Garfield (Eds.), Handbook of psychotherapy and be-havior change (4th edn.), (pp. 595-630). New York: JohnWiley & Sons.

Alexander, J. F. & Parsons, B. V. (1982). Functional familytherapy. Monterey, CA: Brooks/Cole.

American Psychiatric Association (1994). Diagnostic and stat-istical manual of mental disorders (4th edn.). Washington,DC: APA.

Armbruster, P., & Kazdin, A. E. (1994). Attrition in childpsychotherapy. In T. H. Ollendick & R. J. Prinz (Eds.),Advances in clinical child psychology. Vol. 16 (pp. 81-108).New York: Plenum.

Asher, S. R. & Coie, J. D. (Eds.) (1990). Peer rejection inchildhood. New York: Cambridge University Press.

Baer, R. A. & Nietzel, M. T. (1991). Cognitive and behavioraltreatment of impulsivity in children: A meta-analytic reviewof the outcome literature. Journal of Clinical Child Psy-chology, 20, 400-412.

Bank, L., Marlowe, J. H., Reid, J. B., Patterson, G. R. &Weinrott, M. R. (1991). A comparative evaluation of parent-training interventions for families of chronic delinquents.Journal of Abnormal Child Psychology, 19, 15-33.

Bell, R. Q. & Harper, L. (1977). Child ejfects on adults. NewYork: John Wiley & Sons.

Borduin, C. M., Henggeler, S. W., Blaske, D. M. & Stein, R.(1990). Multisystemic treatment of adolescent sexual of-fenders. International Journal of Offender Therapy and Com-parative Criminology, 34, 105-113.

Borduin, C. M., Mann, B. J., Cone, L. T., Henggeler, S. W.,Fucci, B. R., Blaske, D. M. & Williams, R. A. (1995).Multisystemic treatment of serious juvenile offenders: Long-term prevention of criminality and violence. Journal ofConsulting and Clinical Psychology, 63, 569-578.

Brandt, D. E. & Zlotnick, S. J. (1988). The psychology andtreatment of the youthful offender. Springfield, IL: Charles CThomas.

Brunk, M., Henggeler, S. W. & Whelan, J. P. (1987). Acomparison of multisystemic therapy and parent training inthe brief treatment of child abuse and neglect. Journal ofConsulting and Clinical Psychology, 55, 311-318.

Campbell, M., Adams, P. B., Small, A. M., Kafantaris, V.,Silva, R. R., Shell, J., Perry, R. & Overall, J. E. (1995).Lithium in hospitalized aggressive children with conductdisorder: A double-blind and placebo-controlled study.Journal of the American Academy of Child and AdolescentPsychiatry, 34, 445-453.

Campbell, M. & Cueva, J. E. (1995). Psychopharmacology inchild and adolescent psychiatry: A review of the past sevenyears. Part II. Journal ofthe American Academy of Child andAdolescent Psychiatry, 34, 1262-1272.

Capaldi, D. & Patterson, G. R. (1987). An approach to theproblem of recruitment and retention rates for longitudinalresearch. Behavioral Assessment, 9, 169-187.

Cooper, J. O., Heron, T. E. & Heward, W. L. (1987). Appliedbehavior analysis. Columbus, OH: Merrill.

Crick, N. R. & Dodge, K. A. (1994). A review and reformu-lation of social infonnation processing mechanisms inchildren's social adjustment. Psychological Bulletin, 115,74-101.

Cueva, J. E., Overall, J. E., Small, A. M., Armenteros, J. L.,Perry, R. & Campbell, M. (1996). Carbamazepine in ag-gressive children with conduct disorder: A double-blind andplacebo controlled study. Journal of the American Academyof Child and Adolescent Psychiatry, 35, 480-490.

Cunningham, C. E., Bremner, R. & Boyle, M. (1995). Largegroup community-based parenting programs for families of

176 A. E. KAZDIN

preschoolers at risk for disruptive behaviour disorders:Utilization, cost effectiveness, and outcome. Journal of ChildPsychology and Psychiatry, 36, 1141-1159.

Dadds, M. R. & McHugh, T. A. (1992). Social support andtreatment outcome in behavioral family therapy for childconduct problems. Journal of Consulting and Clinical Psy-chology, 60, 252-259.

Dadds, M. R., Schwartz, S. & Sanders, M. R. (1987). Maritaldiscord and treatment outcome in behavioral treatment ofchild conduct disorders. Journal of Consulting and ClinicalPsychology, 55, 396-403.

Dishion, T. J. & Andrews, D. W. (1995). Preventing escalationin problem behaviors with high-risk young adolescents:Immediate and 1 -year outcomes. Journal of Consulting andClinical Psychology, 63, 538-548.

Dishion, T. J. & Patterson, G. R. (1992). Age effects in parenttraining outcomes. Behavior Therapy, 23, 719-729.

Dishion, T. J., Patterson, G. R. & Kavanagh, K. A. (1992). Anexperimental test of the coercion model: Linking theory,measurement, and intervention. In J. McCord & R. E.Tremblay (Eds.), Preventing antisocial behavior (pp. 253-282). New York: Guilford.

Dumas, J. E. (1989). Treating antisocial behavior in children:Child and family approaches. Clinical Psychology Review, 9,197-222.

Dumas, J. E. & Wahler, R. G. (1983). Predictors of treatmentoutcome in parent training: Mother insularity and socio-economic disadvantage. Behavioral Assessment, 5, 301-313.

Duriak, J. A., Fuhrman, T. & Lampman, C. (1991). Effective-ness of cognitive-behavioral therapy for maladapting chil-dren: A meta-analysis. Psychological Bulletin, UO, 204-214.

Falloon, I. R. (1988). Expressed emotion: Current status.Psychological Medicine, 18, 269-274.

Feindler, E. L. & Ecton, R. B. (1986). Adolescent anger control:Cognitive-behavioral techniques. Elmsford, NY: Pergamon.

Feldman, R. A., Caplinger, T. E. & Wodarski, J. S. (1983). TheSt. Louis conundrum: The effective treatment of antisocialyouths. Englewood Cliffs, NJ: Prentice-Hall.

Fergusson, D. M., Horwood, L. J. & Lloyd, M. (1991). Confir-matory factor models of attention deficit and conductdisorder. Journal of Child Psychology and Psychiatry, 32,257-274.

Finch, A. J., Jr., Nelson, W. M. & Ott, E. S. (1983). Cognitive-behavioral procedures with children and adolescents: A prac-tical guide. Needham Heights, MA: Allyn & Bacon.

Forehand, R. & Long, N. (1988). Outpatient treatment oftheacting out child: Procedures, long-term follow-up data, andclinical problems. Advances in Behaviour Research andTherapy, 10, 129-177.

Forehand, R. & McMahon, R.J . (1981). Helping the non-comptiant child: A clinician's guide to parent training. NewYork: Guilford.

Griest, D. L., Forehand, R., Rogers, T., Breiner, J., Furey, W.& Williams, C. A. (1982). Effects of parent enhancementtherapy on the treatment outcome and generalization of aparent training program. Behaviour Research and Therapy,20, 429^36.

Henggeler, S. W. (1994). Treatment manual for family pres-ervation using multisystemic therapy. Charleston, SC: MedicalUniversity of South Carolina, South Carolina Health andHuman Services Finance Commission.

Henggeler, S. W. & Borduin, C. M. (1990). Family therapy andbeyond: A multisystemic approach to teaching the behaviorproblems of children and adolescents. Pacific Grove, CA:Brooks/Cole.

Henggeler, S. W., Melton, G. B. & Smith, L. A. (1992). Familypreservation using multisystemic therapy: An effective alter-native to incarcerating serious juvenile offenders. Journal ofConsulting and Clinical Psychology, 60, 953-961.

Henggeler, S. W., Rodick, J. D., Borduin, C. M., Hanson, C.L., Watson, S. M. & Urey, J. R. (1986). Multisystemictreatment of juvenile offenders: Effects on adolescent be-havior and family interaction. Developmental Psychology, 22,132-141.

Henggeler, S. W., Schoenwald, S. K. &Pickrel, S. A. G. (1995).Multisystemic therapy: Bridging the gap between university-and community-based treatment. Journal of Consulting and

Clinical Psychology, 63, 709-717.Hinshaw, S. P. (1994). Attention deficits and hyperactivity in

children. Thousand Oaks, CA: Sage.Hinshaw, S. P., Lahey, B. B. & Hart, E. L. (1993). Issues of

taxonomy and comorbidity in the development of conductdisorder. Development and Psychopathology, 5, 31-49.

Holmes, D. S. & Urie, R. G. (1975). Effects of preparingchildren for psychotherapy. Journal of Consulting and ClinicalPsychology, 43, 311-318.

Kaminer, Y., Tarter, R. E., Bukstein, O. G. & Kabene, M.(1992). Comparison between treatment completers and non-completers among dually diagnosed substance-abusing ado-lescents. Journal of the American Academy of Child andAdolescent Psychiatry, 31, 1046-1049.

Kazdin, A. E. (1985). Treatment of antisocial behavior in childrenand adolescents. Homewood, IL: Dorsey Press.

Kazdin, A. E. (1988). Child psychotherapy: Developing andidentifying effective treatments. Needham Heights, MA:Allyn & Bacon.

Kazdin, A. E. (1990). Premature termination from treatmentamong children referred for antisocial behavior. Journal ofChild Psychology and Psychiatry, 3, 415^25.

Kazdin, A. E. (1992). Research design in clinical psychology(2nd edn.). Needham Heights, MA: Allyn & Bacon.

Kazdin, A. E. (1993). Treatment of conduct disorder: Progressand directions in psychotherapy research. Development andPsychopathology, 5, 277-310.

Kazdin, A. E. (1994a). Behavior modification in applied settings(5th edn.). Pacific Grove, CA: Brooks/Cole.

Kazdin, A. E. (1994b). Psychotherapy for children and ado-lescents. In A. E. Bergin & S. L. Garfield (Eds.), Handbook ofpsychotherapy and behavior change (4th edn.) (pp. 543-594).New York: Wiley & Sons.

Kazdin, A. E. (1995a). Child, parent, and family dysfunction aspredictors of outcome in cognitive-behavioral treatment ofantisocial children. Behaviour Research and Therapy, 33,271-281.

Kazdin, A. E. (1995b). Conduct disorder in childhood andadolescence (2nd edn). Thousand Oaks, CA: Sage.

Kazdin, A. E. (1996a). Combined and multimodal treatmentsin child and adolescent psychotherapy: Issues, challenges,and research directions. Clinical Psychology: Science andPractice, 3, 69-100.

Kazdin, A. E. (1996b). Dropping out of child psychotherapy:Issues for research and implications for practice. ClinicalChild Psychology and Psychiatry, 1, 133-156.

Kazdin, A. E., Bass, D., Ayers, W. A. & Rodgers, A. (1990a).Empirical and clinical focus of child and adolescent psycho-therapy research. Journal of Consulting and Clinical Psy-chology, 58, 729-740.

Kazdin, A. E., Bass, D., Siegel, T. & Thomas, C. (1989).Cognitive-behavioral treatment and relationship therapy inthe treatment of children referred for antisocial behavior.Journal of Consulting and Clinical Psychology, 57, 522-535.

Kazdin, A. E. & Crowley, M. (in press). Moderators oftreatment outcome in cognitively based treatment of anti-social children. Cognitive Therapy and Research.

Kazdin, A. E., Esveldt-Dawson, IC., French, N. H. & Unis,A. S. (1987a). Problem-solving skills training and relationshiptherapy in the treatment of antisocial child behavior. Journalof Consulting and Clinical Psychology, 55, 76-85.

TREATMENT OF CONDUCT DISORDER 177

Kazdin, A, E,, Esveldt-Dawson, K,, French, N. H, & Unis, A,S, (1987b), The effects of parent management training andproblem-solving skills training combined in the treatment ofantisocial child behavior. Journal of the American Academy ofChild and Adolescent Psychiatry, 26, 416-424,

Kazdin, A, E,, Mazurick, J, L. & Bass, D, (1993), Risk forattrition in treatment of antisocial children and families.Journal of Clinical Child Psychology, 22, 2-16.

Kazdin, A, E,, Mazurick, J, L, & Siegel, T, C, (1994). Treatmentoutcome among children with externalizing disorder whoterminate prematurely versus those who complete psycho-therapy. Journal of the American Academy of Child andAdolescent Psychiatry, 33, 549-557,

Kazdin, A, E,, Siegel, T. C. & Bass, D. (1990b), Drawing uponclinical practice to inform research on child and adolescentpsychotherapy: A survey of practitioners. Professional Psy-chology: Research and Practice, 21, 189-198,

Kazdin, A, E,, Siegel, T, & Bass, D, (1992), Cognitive problem-solving skills training and parent management training in thetreatment of antisocial behavior in children. Journal ofConsulting and Clinical Psychology, 60, Il>'i-1A1.

Kendall, P, C, (Ed,) (1991), Child and adolescent therapy:Cognitive-behavioral procedures. New York: Guilford,

Kiesler, D, J, (1971), Experimental designs in psychotherapyresearch. In A. E, Bergin & S, L. Garfield (Eds,), Handbookof psychotherapy and behavior change: An empirical analysis(pp, 36-74), New York: Wiley,

Kupfer, D, J,, Frank, E,, Perel, J. M., Comes, C , Mallinger, A,G,, Thase, M, E,, McEachran, A. B, & Grochocinski, V. J,(1992), Five-year outcome for maintenance therapies inrecurrent depression. Archives of General Psychiatry, 49,169-112,.

Lochman, J. E. & Dodge, K. A, (1994), Social-cognitive pro-cesses of severely violent, moderately aggressive, and non-aggressive boys. Journal of Consulting and Clinical Psy-chology, 62, 366-374,

Long, P,, Forehand, R,, Wierson, M, & Morgan, A. (1994).Does parent training with young noncompliant children havelong-term effects? Behaviour Research and Therapy, 32,101-107,

Lundman, R, J, (1984), Prevention and control of juveniledelinquency. New York: Oxford University Press.

Lytton, H. (1990). Child and parent effects in boys' conductdisorder: A reinterpretation. Developmental Psychology, 26,683-697,

Mann, B, J,, Borduin, C, M,, Henggeler, S, W, & Blaske, D, M.(1990), An investigation of systemic conceptualizations ofparent-child coalitions and symptom change. Journal ofConsulting and Clinical Psychology, 58, 336-344,

McMahon, R, J,, Forehand, R,, Griest, D, L, & Wells, K, C,(1981), Who drops out of treatment during parent behavioraltraining? Behavioral Counseling Quarterly, 1, 79-85.

McMahon, R. J, & Wells, K, C, (1989), Conduct disorders. InE, J, Mash & R, A, Barkley (Eds,), Treatment of childhooddisorders (pp, 73-132), New York: Guilford Press,

Miller, G, E, & Prinz, R, J, (1990), Enhancement of sociallearning family interventions for child conduct disorder.Psychological Bulletin, 108, 291-307,

Moffitt, T, E, (1993), The neuropsychology of conduct prob-lems. Development and Psychopathology, 5, 135-151,

Morris, S, M,, Alexander, J, F, & Turner, C, W, (1991), Doreattributions reduce blame? Journal of Family Psychology,5, 192-203,

Newberry, A, M,, Alexander, J, F, & Turner, C, W, (1991),Gender as a process variable in family therapy. Journal ofFamily Psychology, 5, 158-175,

O'Donnell, C, R. (1992), The interplay of theory and practice indelinquency prevention: From behavior modification toactivity settings. In J, McCord & R, E, Tremblay (Eds,),

Preventing antisocial behavior (pp, 209-232), New York:Guilford,

Offord, D, R,, Boyle, M, H, & Racine, Y. A. (1991), Theepidemiology of antisocial behavior. In D, J. Pepler & K, H.Rubin (Eds.), The development and treatment of childhoodaggression (pp, 31-54), Hillsdale, NJ: Erlbaum,

Patterson, G. R, (1982), Coercive family process. Eugene, OR:Castalia,

Patterson, G, R., Capaldi, D. & Bank, L, (1991), An earlystarter model for predicting delinquency. In D, J, Pepler & K,H, Rubin (Eds,), The development and treatment of childhoodaggression (pp. 139-168). Hillsdale, NJ: Erlbaum,

Patterson, G, R, & Chamberlain, P, (1994), A functionalanalysis of resistance during parent training therapy. ClinicalPsychology: Science and Practice, 1, 53-70,

Patterson, G, R,, DeBaryshe, B, D, & Ramsey, E. (1989). Adevelopmental perspective on antisocial behavior. AmericanPsychologist, 44, 329-335.

Patterson, G, R,, Dishion, T, J, & Chamberlain, P, (1993).Outcomes and methodological issues relating to treatment ofantisocial children. In T, R. Giles (Ed,), Handbook of effectivepsychotherapy (pp. 43-87), New York: Plenum,

Patterson, G. R., Reid, J. B. & Dishion, T, J, (1992), Antisocialboys. Eugene, OR: Castalia,

Pepler, D, J, & Rubin, K, H. (Eds.) (1991), The development andtreatment of childhood aggression. Hillsdale, NJ: Erlbaum,

Prinz, R, J, & Miller, G, E, (1994), Family-based treatment forchildhood antisocial behavior: Experimental influences ondropout and engagement. Journal of Consulting and ClinicalPsychology, 62, 645-650,

Robins, L, N, (1981). Epidemiological approaches to naturalhistory research: Antisocial disorders in children. Journal ofthe American Academy of Child Psychiatry, 20, 566-680.

Robins, L, N. (1991). Conduct disorder. Journal of ChildPsychology and Psychiatry, 32, 193-212.

Rubin, K, H,, Bream, L, A, & Rose-Krasnor, L, (1991), Socialproblem solving and aggression in childhood. In D. J, Pepler& K, H, Rubin (Eds,), The development and treatment ofchildhood aggression (pp, 219-248), Hillsdale, NJ: Erlbaum,

Rutter, M. & Giller, H, (1983), Juvenile delinquency: Trends andperspectives. New York: Penguin Books,

Sanders, M, R, & Dadds, M, R, (1993), Behavioral familyintervention. Needham Heights, MA: Allyn & Bacon,

Santisteban, D.A, Szapocznik, J., Perez-Vidal, A,, Kurtines, W,H,, Murray, E, J, & LaPerriere, A, (1996), Efficacy ofintervention for engaging youth and families into treatmentand some variables that may contribute to differential effec-tiveness. Journal of Family Psychology, 10, 35^4,

Shirk, S, R, (Ed,) (1988). Cognitive development and childpsychotherapy. New York: Plenum.

Shure, M. B, (1992), / can problem solve (ICPS): An inter-personal cognitive problem solving program. Champaign, IL:Research Press.

Spivack, G. & Shure, M, B, (1982). The cognition of socialadjustment: Interpersonal cognitive problem solving think-ing. In B. B, Lahey & A, E, Kazdin (Eds,), Advances inclinical child psychology. Vol. 5 (pp, 323-372), New York:Plenum,

Stewart, J, T,, Myers, W, C , Burket, R, C, & Lyles, W. B.(1990), A review of the psychopharmacology of aggression inchildren and adolescents. Journal of the American Academy ofChild and Adolescent Psychiatry, 29, 269-277,

Szapocznik, J,, Perez-Vidal, A,, Brickman, A,, Foote, F, H,,Santisteban, D,, Hervis, O, & Kurtines, W, H. (1988),Engaging adolescent drug abusers and their families intotreatment: A strategic structural systems approach. Journalof Consulting and Clinical Psychology, 56, 552-557.

Szapocznik, J., Rio, A., Murray, E,, Cohen, R., Scopetta, M,,Rivas-Vasquez, A,, Hervis, O,, Posada, V. & Kurtines, W,

178 A. E. KAZDIN

(1989). Structural family versus psychodynamic child therapyfor problematic Hispanic boys. Journal of Consulting andClinical Psychology, 57, 571-578.

United States Congress, Office of Technology Assessment.(1991). Adolescent health. (OTA-H-468). Washington, DC:U.S. Government Printing Office.

Wahler, R. G. & Dumas, J. E. (1986). Maintenance factors incoercive mother-child interactions: The compliance andpredictability hypotheses. Journal of Applied Behavior Analy-sis, 19, 13-22.

Walker, J. L., Lahey, B. B., Russo, M. F., Christ, M. A. G.,McBurnett, K., Loeher, R., Stouthamer-Loeher, M. & Green,S. M. (1991). Anxiety, inhibition, and conduct disorder inchildren: I. Relation to social impairment. Journal of theAmerican Academy of Child and Adolescent Psychiatry, 30,187-191.

Wehster-Stratton, C. (1985). Predictors of treatment outcomein parent training for conduct disordered children. BehaviorTherapy, 16, 223-243.

Webster-Stratton, C. (1994). Advancing videotape parent train-ing : A comparison study. Journal of Consulting and ClinicalPsychology, 62, 583-593.

Webster-Stratton, C , Hollinsworth, T. & Kolpacoff, M. (1989).The long-term effectiveness and clinical significance of threecost-effective training programs for families with conduct-prohlem children. Journal of Consulting and Clinical Psy-chology, 57, 550-553.

Wierzbicki, M. & Pekarik, G. (1993). A meta-analysis ofpsychotherapy dropout. Professional Psychology: Researchand Practice, 24, 190-195.

Weisz, J. R., Walter, B. R., Weiss, B., Fernandez, G. A. &Mikow, V. A. (1990). Arrests among emotionally disturhedviolent and assaultive individuals following minimal versuslengthy intervention through North Carolina's Willie M.Program. Journal of Consulting and Clinical Psychology, 58,720-728.

Accepted manuscript received 19 May 1996