رفتاردرمانd شناختی برای خشم در کودکان و نوجوانان: یک متاآنالیز

The meta-analysis of the treatment outcome studies of cognitive-behavioral therapy (CBT) for anger-related problems in children and adolescents included 21 published and 19 unpublished reports. The mean effect size (Cohen's d=0.67) was in the medium range and consistent with the effects of psychotherapy with children in general. The differential effects of skills training, problem solving, affective education, and multimodal interventions (d=0.79, 0.67, 0.36, and 0.74, respectively) were variable although also generally in the medium range. Skills training and multimodal treatments were more effective in reducing aggressive behavior and improving social skills. However, problem-solving treatments were more effective in reducing subjective anger experiences. Modeling, feedback, and homework techniques were positively related to the magnitude of effect size.

Anger-related problems, such as oppositional behavior, hostility, and aggression, are some of the main reasons that children and adolescents are referred for counseling or psychotherapy Abikoff & Klein, 1992 and Armbruster et al., 2001. While anger-related problems constitute the central feature of disruptive behavior disorders and are frequently associated features of attention-deficit hyperactivity disorder (American Psychiatric Association, 1994), they are often present in other childhood disorders. Inspection of the DSM-IV disorders applicable to youth reveals several diagnostic criteria, associated features, and descriptors that are relevant to anger. Irritability is a prominent feature of all major mood disorders, including bipolar disorders and depressive disorders. In adjustment disorders involving disturbance of emotions or conduct, there often are violation of the rights of others, aggressive behavior, and persistent anger. Aggressiveness, poor impulse control, and intense anger and hostility are, likewise, characteristics of a broad range of disorders involving abuse or withdrawal from alcohol or other drugs. Intermittent explosive disorder is defined primarily by discrete episodes of loss of control of aggressive behavior. Finally, Tourette's disorder and obsessive-compulsive disorder in children may cooccur with temper tantrums and oppositional behavior. 1.1. Phenomenology and elements of the anger construct Several models of anger were considered to provide a conceptual framework for this meta-analysis. Novaco (1975) proposed a model of anger, which includes subjective emotional states, environmental circumstances, physiological arousal, cognitions of antagonism, and corresponding behavioral reactions. The subjective affect is determined by cognitive labeling of physiological arousal as “being angry.” This cognitive labeling is a highly automatic process, which is associated with an inclination to act in a confrontational manner toward the source of provocation. This action impulse is regulated by internal and external mechanisms of control, which may be overridden by the intensity of any one of the elements of anger. Spielberger (1988) proposed a factor-analytical model of anger that distinguished between anger experience and anger expression. Within this model, anger experience is viewed as a subjective experience varying in duration and intensity. Anger expression is viewed as an individual's tendency to act on anger by showing it outwardly, suppressing it, or actively coping with it. However, Spielberger et al. (1983) also suggested that there are unclear boundaries among the related concepts of anger, hostility, and aggression and that the three can be integrated into a collective “AHA syndrome.” Within this syndrome, anger refers to emotional states, hostility refers to antagonistic beliefs, and aggression refers to overt harmful behavior. Several social-cognitive models have detailed cognitive processes that may be related to anger and aggression. These models stem from the original social learning formulations by Bandura (1973) as well as models of problem solving (d'Zurilla & Goldfried, 1971) and causal attribution (Kelley, 1972). The social information processing model developed by Dodge (1980) postulated a five-step sequential model of cognitive processes: encoding of social cues, interpretation of cues, response search, response decision, and enactment of behavior. Disruption in any of these processes can lead to anger and aggressive behavior. Kendall (1991) made a distinction between cognitive deficiencies and cognitive distortions. Deficiencies refer to the absence of thinking, such as not thinking about the consequences of one's behavior, and distortions, such as a hostile attribution bias, refer to the faulty processing of social information. Cognitive deficiencies require interventions that enrich the repertoire of cognitive and behavioral skills, whereas cognitive distortions require modification of already existing cognitive and behavioral patterns. In this meta-analysis, the construct of anger was used as one of the selection criteria for the outcome studies. Anger was defined as a subjective, negatively felt state associated with cognitive deficits and distortions and maladaptive behaviors Kassinove & Sukhodolsky, 1995 and Martin et al., 2000. The phenomenology of anger includes emotional experiences, varying from annoyance to rage, behavioral patterns, varying from social withdrawal to physical aggression, and cognitive phenomena, such as attributions of blame and mental rumination. Studies of cognitive-behavioral therapy (CBT) for anger-related problems in children and adolescents (herein called “children”) were considered for inclusion. 1.2. CBT for anger in children For the purposes of this study, CBT was defined as a class of child-focused treatments that target covert and overt behaviors to accomplish improvement in symptoms and functioning Beidel & Turner, 1986 and Spiegler & Guevremont, 1993. Therefore, interventions that are delivered to adults (e.g., parent management training) and interventions focused on altering environmental contingencies (e.g., multisystemic therapy) to improve child functioning were not considered. The rationale for conducting a meta-analysis of CBT for anger in children was twofold. First, therapies based on stress inoculation (Meichenbaum & Cameron, 1973) and arousal reduction (Suinn & Richardson, 1971) models have been a predominant form of treatment for general anger since the 1970s. Second, several treatments for anger in children are based on social-cognitive theory and use cognitive-behavioral procedures. Within this tradition, possible cognitive mediators of aggression such as attributional processes (Hudley & Graham, 1993), biased perception of social cues (Dodge & Crick, 1990), and deficient social problem-solving skills (Lochman, Meyer, Rabiner, & White, 1991) are targeted for intervention. A recent review suggested that CBT is generally effective for the treatment of anger (Beck & Fernandez, 1998); however, the differential effects of CBT subtypes have not been investigated. Although united by the similar theoretical backgrounds, cognitive-behavioral treatments vary in terms of specific techniques and target symptoms. Therefore, we distinguished among the types of CBT based on the predominant therapeutic techniques and on the targeted element of anger construct. We also adapted Kendall's (1993) classification of cognitive-behavioral procedures for youth (modeling, building cognitive coping skills, using rewards to modify behavior, rehearsing appropriate behavior, affective education, and training tasks) to identify the categories that were used in this meta-analysis. Considering both treatment targets and therapeutic procedures, four categories of CBT were identified. (1) Skills development category: this included treatments that targeted overt anger expression and used modeling and behavioral rehearsal to develop appropriate social behaviors. (2) Affective education category: this included treatments that focused on covert anger experience and included techniques of emotion identification, self-monitoring of anger arousal, and relaxation. (3) Problem-solving category: this included treatments that targeted cognitive deficits and distortions and used techniques such as attributional training, self-instruction, and consequential thinking. (4) Eclectic or multimodal treatment category was used to incorporate studies that use multiple procedures and targeted two or more components of anger. 1.3. Moderating and mediating variables Identification of factors that predict and influence children's response to therapy is an important task of psychotherapy research (Kazdin & Weisz, 1998). However, according to Beutler (1991), “there are nearly 1.5 million potential combinations of therapy, therapist, phase, and patient types that must be studied to rule out relevant differences among treatment types” (p.227). Thus, the study of moderating and mediating variables of psychotherapy outcomes becomes a challenge. Compared with treatment outcome studies, meta-analysis provides more statistical power to investigate some of these combinations. We were interested in investigating age, gender, and problem severity as possible moderators of treatment effects. Durlak, Fuhrman, and Lampman (1991) demonstrated that the cognitive-developmental level, as derived from age, was the only significant moderator of the effectiveness of CBT. The effect size (d=0.92) for children presumably functioning at the formal operational level (ages 11–13) was almost twice that (d=0.56) for children at less advanced cognitive stages (ages 5–11). In the second meta-analysis, the severity of impairment was a significant predictor of psychotherapy outcome, but only when specific symptoms were targeted by intervention (Durlak, Wells, Cotten, & Johnson, 1995). Therapy outcomes may vary as a function of a variety of factors that unfold during treatment. Some of these factors include treatment duration (e.g., brief vs. long-term), format of treatment delivery (e.g., group vs. individual), treatment setting (e.g., clinic vs. school), and therapist characteristics (e.g., experience). While treatment duration can be easily conceptualized and measured, its study has been rarely a focus of independent investigations (Koss & Shiang, 1994). Early clinical reports suggested a linear relationship between number of sessions and improvement, which usually occurs within the first 20 sessions (Strassberg, Anchor, Cunningham, & Elkins, 1977). The more recent “dose–response” model (Howard, Lueger, Martinovich, & Lutz, 1999) suggested that the rate of improvement is the highest earlier in treatment, and it diminishes as the number of sessions increases. Regarding group versus individual formats of treatment delivery, two studies directly compared these formats for the treatment of children's anger Kendall & Zupan, 1981 and Shechtman & Ben-David, 1999, and no significant differences were found. However, a concern arose that group therapy may be detrimental to delinquent youth because it provides opportunities for forming delinquent groups and socialization of antisocial behavior Arnold & Hughes, 1999 and Dishion et al., 1999. The variable of treatment setting is relevant to the understanding of generalizability of treatment effects and exportability of treatments. Treatments evaluated in clinical settings usually have more modest results than those evaluated in research settings (Kazdin, 1995). The role of therapist's experience has been a controversial topic in psychotherapy outcome research (Beutler, Machado, & Neufeld, 1994). Specifically, therapist experience is usually operationalized as the amount or type of training as opposed to the duration of direct experience in using specific treatment for specific population. Schneider (1992) abandoned the attempt to code therapists' experience level and classified therapists' characteristics into two categories: teacher and research assistant/psychology student. These characteristics were used in a meta-analysis of social skills training interventions for children and yielded no significant association with the magnitude of effect size. In the analysis of methodological issues in child psychotherapy research, Durlak et al. (1995) suggested five levels of the therapist experience variable—professional, graduate student, paraprofessional, mixed, and unknown—which were used in this study. 1.4. Measurement characteristics Low correlations between different informants have been noted in evaluating children's behavior problems (Garrison & Earls, 1985). Achenbach, McConaughy, and Howell (1987) distinguished among six groups of informants: mental health worker, observer, parent, peer, self, and teacher. The degrees of association between informants in the same category were high (Pearson r of >.50), while the degrees of association between different categories were relatively small (Pearson r's of .10–.29). Different informants, however, can validly contribute different information about samples of behavior in different situations (e.g., parents at home, teachers at school, and clinicians in the clinic). Thus, it is essential to preserve the contributions of different informants in the assessment of dependent variables in outcome studies. A qualitative review of the studies selected for the present meta-analysis suggested six categories for the source of information variable: self, observation, life record (archival data), parent, teacher, and peers. The variability among the dependent measures used by researchers to evaluate the outcomes of their treatments leaves the problem of grouping these measures according to the judgement of the reviewer. Guided by our interest in anger and anger-related behavioral problems, we grouped the outcome measures used in individual studies into five domains. The anger experience domain included self-reported measures of anger intensity and arousal. The physical aggression domain included measures of aggressive and disruptive behavior that were completed by various informants. The social-cognitive domain included various paper-and-pencil tasks of beliefs about aggression, hostile attribution bias, and decision making. The self-control domain included measures of self-monitoring and self-regulation that were based either on self-reports or on ratings. The social skills domain included either observational or other-report measures of social competencies. 1.5. Objectives of the study There were three main objectives in this meta-analysis: (1) to evaluate the overall effect size of CBT for anger-related problems in children, (2) to compare the effect sizes of skills development, affective education, problem solving, and multimodal interventions, and (3) to explore the effects of CBT across the domains of outcome measures and the categories of informants. In addition, the relationships between the magnitude of treatment effects and the mediating and moderating variables were explored.