| ||Literature Coverage Dates||Number of Studies||Number of Study Participants|
|Meta-Analysis 1||1974 - 1997||40||1953|
|Meta-Analysis 2||1981 - 2010||6||164|
Sukhodolsky, Kassinove, and Gorman (2004) conducted a meta-analysis to examine the effects of cognitive behavioral therapy (CBT) for anger-related problems in children and adolescents. The researchers conducted a literature search via PsycLit, Medline, and Dissertation Abstracts International. They retrieved article abstracts by cross- referencing the following terms: “anger”, “aggression”, “oppositional behavior”, and “antisocial behavior” with “children”, “adolescents”, “treatment”, “therapy”, and “counseling”. Next, they manually examined references of individual outcome articles and meta- analyses of psychotherapy with children to find relevant titles.
Studies were eligible for inclusion if 1) a form of CBT was compared with a no-treatment or attention control condition; 2) treatment targets were explicitly stated and included one or more of the following: anger reduction, reduction of aggressive or antisocial behavior, improvement of anger-related, social-cognitive deficits, improvement of self-regulation or self-control, and improvement of social skills; 3) at least one outcome measure of anger or aggression was included; 4) participants were children and/or adolescents from 6 to 18 years of age; 5) study results were expressed numerically in a way that permitted the computation of effect size; and 6) studies were completed between 1974 and 1997 and reported in English .
Forty studies met the inclusion criteria. Of the 40 studies, 21 were published, and 19 were unpublished. Across these studies, 51 treatment versus control comparisons were used in the meta-analysis. Of these 51 comparisons, 41 were from studies that used random assignment, and 10 were from studies that did not use random assignment. A total of 1,953 children or adolescents participated in the studies. The average age of participants was 12.5, although mean age for the treatment groups ranged from 7 to 17.2 years old. On average, for each comparison, 82 percent of participants were male. Of the total sample, 41 percent of participants were in the moderate range of problem severity, 39 percent were in the mild range, and20 percent were in the severe range of problem severity. Treatment was delivered in various settings, including schools, outpatient centers, inpatient centers, and correctional facilities.
The 40 studies yielded 173 effect sizes. Instead of weighting, the researchers calculated three sets of effects: 1) overall effect per study or comparison, 2) effects per domain of measurement, and 3) effects per source of information. Effects were reported as Cohen’s d
. The number of effects by outcome varied: the aggression outcome was calculated with 36 effect sizes, the anger experience outcome was calculated with 29 effect sizes, the self-control outcome was calculated with 8 effect sizes, the problem-solving outcome was calculated with 11 effect sizes, and the social skills outcome was calculated with 20 effect sizes. The researchers did not indicate what type of model they used for the analyses.Meta-Analysis 2
Hoogsteder and colleagues (2015) conducted a meta-analysis of cognitive behavioral treatment (CBT) for anger-related problems in children and adolescents. The researchers systematically searched the following electronic databases: PsycINFO, Medline, ERIC, Picarta, International Bibliography of the Social Sciences, Adlib, ScienceDirect, SpringerLink, ProQuest Dissertation Abstracts, and Google Scholar by cross-referencing the following keywords: “aggression treatment”, “anger-management training”, “chronic aggression”, “cognitive-behavior therapy”, and “externalizing behavior” with “adolescents”, “youth”, “juvenile”, “inpatient”, “incarcerated”, and “individual”. They then inspected all the references and citations of the articles found and inspected the reference sections of relevant systematic reviews and meta-analyses to find more studies that had not yet been included. Finally, they approached several researchers to obtain unpublished studies.
Studies published between 1980 through 2011 were eligible for inclusion if they met the following criteria: 1) addressed the effectiveness of treatments for adolescents with severe aggressive problems, often accompanied with conduct disorder (CD); 2) examined interventions that were individually oriented, which means that the intervention contained at least an individual component, possibly in combination with group and/or family therapy; 3) included CBT elements such as anger-management training, skills training, and cognitive restructuring; 4) provided posttest scores and a control group; 5) included a control group that received group therapy and/or individual treatment with no CBT elements; 6) included adolescents aged 12–18 years; and 7) provided the necessary data for the calculation of effect sizes. Studies were excluded if they specifically focused on prevention-based treatments designed for youth with conduct problems.
Overall six quasi-experimental studies were identified for inclusion in the meta-analysis. A total of 164 adolescents between the ages of 12 and 18 participated in the studies, which were conducted in both inpatient and outpatient settings. The average age was 15.8 years old, and approximately 63 percent of participants were male. In regard to race/ethnicity, 41 percent were black, 24 percent were white, and 5 percent were Latino; however two of the studies did not provide information on cultural background. Approximately 74 percent of participants (n
= 153) had a mental disorder, 64 percent were diagnosed with conduct disorder, 26 percent with oppositional de?ant disorder, 47 percent with a posttraumatic stress disorder (PTSD), and 30 percent had comorbid diagnoses. In two of the studies, the control groups received treatment as usual, which consisted of individual therapy with no CBT elements. In four of the studies, control groups received group therapy with elements of CBT.
The six studies yielded 13 effect sizes. Cohen’s d
was calculated using the mean scores and standard deviations differences between the experimental group and the control group. Cohen’s d was adjusted for pretest group differences in the outcome variables. The researchers used a multilevel random effects model to calculate the combined effect sizes. The overall effect size was homogenous; therefore, the researchers did not conduct analyses to examine if moderators, such as gender or type of CBT intervention, had any influence on the outcomes.