Eur Child Adolesc Psychiatry DOI 10.1007/s00787-014-0592-1
REVIEW
Treatment moderators of cognitive behavior therapy to reduce aggressive behavior: a meta-analysis Kirsten C. Smeets • Anouk A. M. Leeijen • Marie¨t J. van der Molen • Floor E. Scheepers Jan K. Buitelaar • Nanda N. J. Rommelse
•
Received: 5 November 2013 / Accepted: 25 July 2014 Ó Springer-Verlag Berlin Heidelberg 2014
Abstract Maladaptive aggression in adolescents is an increasing public health concern. Cognitive Behavior Therapy (CBT) is one of the most common and promising treatments of aggression. However, there is a lack of information on predictors of treatment response regarding CBT. Therefore, a meta-analysis was performed examining the role of predictors on treatment response of CBT. Twenty-five studies were evaluated (including 2,302 participants; 1,580 boys and 722 girls), and retrieved through searches on PubMed, PsycINFO and EMBASE. Effect sizes were calculated for studies that met inclusion criteria. Study population differences and specific CBT characteristics were examined for their explanatory power. There was substantial variation across studies in design and outcome variables. The meta-analysis showed a medium treatment effect for CBT to reduce aggression (Cohen’d = 0.50). No predictors of treatment response were found in the meta-analysis. Only two studies did Trial registration: ISRCTN21681959.
Electronic supplementary material The online version of this article (doi:10.1007/s00787-014-0592-1) contains supplementary material, which is available to authorized users. K. C. Smeets (&) J. K. Buitelaar N. N. J. Rommelse Karakter Child and Adolescent Psychiatry University Centre, Reinier Postlaan 12, 6525 GC Nijmegen, The Netherlands e-mail: [email protected] A. A. M. Leeijen Department of Health Sciences, VU University Amsterdam, Amsterdam, The Netherlands M. J. van der Molen Faculty of Psychology and Education and the EMGO Institute for Health and Care Research, VU University Amsterdam, Amsterdam, The Netherlands
examine whether proactive versus reactive aggression could be a moderator of treatment outcome, and no effect was found of this subtyping of aggression. These study results suggest that CBT is effective in reducing maladaptive aggression. Furthermore, treatment setting and duration did not seem to influence treatment effect, which shows the need for development of more cost-effective and less-invasive interventions. More research is needed on moderators of outcome of CBT, including proactive versus reactive aggression. This requires better standardization of design, predictors, and outcome measures across studies. Keywords Aggression Cognitive behavior therapy Meta-analysis Effect size
Introduction Maladaptive aggressive behavior is common in adolescents and one of the most prevalent reasons for referral to mental health services [1, 2]. Aggression can be defined as behavior directed at an object, human or animal, through which the object, human or animal experiences harm or F. E. Scheepers Department of Psychiatry, Rudolf Magnus Institute of Neuroscience, UMC Utrecht, Utrecht, The Netherlands J. K. Buitelaar Department of Cognitive Neuroscience, Donders Institute for Brain, Cognition and Behavior, Radboud University Medical Center, Nijmegen, The Netherlands N. N. J. Rommelse Department of Psychiatry, Radboud University Medical Center, Nijmegen, The Netherlands
123
Eur Child Adolesc Psychiatry
damage [3]. Severe maladaptive aggression in adolescents is accompanied by substantial community costs [4]. One of the reasons is that aggression often demands immediate attention, because aggressive behavior leads to a variety of problems in different life domains, like being expelled from school or contact with police or justice. Children and adolescents with symptoms of aggressive behavior and disruptive behavior problems are often diagnosed with Oppositional Defiant Disorder (ODD) or Conduct Disorder (CD) [5]. Also, aggression in childhood or adolescents persists often in severe aggression in adulthood [6, 7]. A literature review shows that 86 % of the children with CD at the age of seven still have conduct problems at the age of fifteen. Furthermore, 5–10 % of all children with antisocial behavior problems continue to present with serious problems in adulthood, and 50–60 % of the crimes in youth are committed by this persistent group [8]. To reduce these costs and public health concerns, effective treatment of aggression is needed. Both pharmacologic and non-pharmacologic approaches have been applied to treat aggression in adolescents [9–13]. Atypical antipsychotics, of which Risperidone is the most commonly studied, presynaptic alpha-adrenergic agonists and psychostimulants have shown to be effective in reducing aggression in adolescents [9, 10, 13]. Furthermore, Cognitive Behavioral Therapies (CBT) such as anger management, multimodal interventions and skills training have shown to be effective in reducing aggression [14]. Several treatment guidelines recommend psychosocial treatment as the first option of treatment of aggression in children and adolescents, with consideration of pharmacologic treatment in patients who do not respond (no improvement of their aggressive behavior) to the psychosocial treatment [5]. However, there is a lack of information on predictors of treatment success regarding CBT [14, 15]. More specific knowledge of the characteristics of participants that respond to interventions could help clinicians identify the best treatment of aggression for individual patients. Characteristics of the participants (e.g., age, gender), treatment (e.g., duration, format or setting) or quality of the study design (e.g., open versus blinded raters) can be considered as possible moderators of treatment effect [16]. For example, CBT in older children seems to be more effective than CBT in younger children. Also, the number of sessions (around 20 sessions) is positively associated with the outcome of the treatment [14]. In addition, studies reporting clinical diagnoses of the participants have larger effect sizes than studies that do not report diagnoses [15]. Little attention has been paid so far to the distinction between reactive and proactive aggression as moderators of treatment effect. There is a body of evidence on the distinction between the two types of aggression, and treatment regarding proactive and reactive
123
aggression might be different [17, 18]. Reactive aggression is also commonly known as ‘impulsive’, ‘hot-blooded’, ‘affective’ or ‘defensive’ aggression and refers to emotionally charged responses to provocations or frustration [19, 20]. Proactive aggression is also commonly called ‘instrumental’, ‘cold-blooded’ or ‘predatory’ aggression and refers to conscious and planned acts that are often enacted on the basis of egocentric motives for personal gains [21–23]. The distinction between these types of aggression may be relevant in predicting treatment success. Reactive aggressors seem to benefit more from psychosocial interventions than proactive aggressors [24, 25]. Children and adolescents with severe problems of CD and proactive aggression have found to be more resistant for treatment and often do not clinically improve after following psychosocial treatment [25–27]. On the other hand, proactive and reactive aggression are often highly correlated [28, 29], and there is ongoing debate whether proactive and reactive aggression need to be distinguished or are components of one overarching construct [30]. Therefore, a meta-analysis was conducted of CBT studies that aimed to reduce aggression in children and adolescents, including their predictors of treatment outcome. Besides the common predictors such as gender, age, diagnosis, setting or treatment format, proactive and reactive aggression were also investigated as potential predictors of success regarding CBT.
Methods A systematic and comprehensive search for studies in English for the period January 2000 until April 2013 was conducted. Since previous meta-analyses regarding CBT of aggression in children and adolescents were based on literature of the period between 1968 and 2005 [14, 15], this meta-analysis was based on the most recent literature. PubMed, PsycINFO and EMBASE were searched for cognitive behavioral therapies used to treat maladaptive aggression. The search included, among others, the following keywords: aggression, cognitive behavioral therapy and randomized controlled trial (see Supplementary material Appendix I for the complete list of search terms). In addition to searching these databases, reference lists from all retrieved articles and relevant published reviews (i.e., [14, 15]) were hand-searched to identify additional related publications (see Supplementary material Appendix II for a flow chart of the search strategy). Within each individual study, outcome measures relevant to maladaptive aggression were selected during the coding stage. Only one outcome measurement per study was selected, to secure independency of the data. If more outcome measurements were present in the study, the most commonly
Eur Child Adolesc Psychiatry
used (i.e., CBCL questionnaire) measurement was chosen. Next, the effect sizes of the treatment were calculated; see below for details [31]. Studies were included if: 1.
2.
3.
4.
Cognitive behavioral strategies had been used such as coaching and modeling, and if participants’ behavior and thinking styles were addressed using techniques such as anger management, social skills training or assertiveness training [15, 32]. An experimental or quasi-experimental design had been used that assigned participants to either CBT intervention or control group (thus only Randomized Controlled Trials were included). Acceptable comparison conditions were placebo, waiting list, or treatment as usual conditions. They reported at least one quantitative measure on aggressive behavior (e.g., rating scale or observation scale, pre- and post-measurements) subsequent to treatment as an outcome variable. The CBT treatment included participants under the age of 23 with maladaptive aggression problems.
Studies were excluded if no intervention or control treatment was defined. Studies where part of the participants received medication treatment were not excluded. The following possible moderators of treatment outcome were extracted and analyzed: age, gender, diagnosis of ODD/CD, duration of treatment, setting of treatment and type of CBT treatment (family, group or individual). Furthermore, we coded whether the study outcome was assessed by non-blinded raters who were aware of treatment allocation, or by blinded raters. The quality of the study was evaluated based on the Jadad scale (rating regarding quality of randomization, blinding and treatment of missing data) [33]. These variables regarding study quality were also taken into account as possible moderators of treatment effect [16]. Statistical analyses Within each individual study, outcome measures relevant to maladaptive aggression were selected during the coding stage. First, the studies were selected by one of the authors. In cases of doubt and after the first selection was made, another author coded the studies as well (double-coding). The pretest–posttest-control group design was used to analyze the treatment effects. Effect sizes were calculated using the dppc2 developed by Morris [31]: dppc2 ¼ Cp
ðMPost;T MPre;T Þ ðMPost;C MPre;C Þ SDPre
where Mpost, T referred to the mean treatment population aggression score post test, Mpre, T referred to the mean treatment population aggression score pre-test, Mpost, C
referred to the mean control population aggression score post test and Mpre, C referred to the mean control population aggression score pre-test. SDpre, referred to the standard deviation of the pre-test and Cp is the sample size bias correction, based on the n in the treatment group (Nt) and the n in the control group (Nc) [31]: Cp¼1
3 1 4ðNt þ Nc 2Þ
An effect size of d = 0.20 indicates a small effect, an effect size of d = 0.50 denotes a medium effect and an effect size of d = 0.80 represents a large effect. Furthermore, negative effect sizes assume favorable effects in control groups [34]. Also, heterogeneity was calculated in Review Manager 5.1 (RevMan) [35]. A meta-regression analysis was conducted to determine the different possible moderators of effect (sensitivity analysis). The effect of moderators on the treatment effect sizes was analyzed by a t test or linear regression analysis for continuous variables and Chi-square test for categorical variables, using SPSS 20.0.
Results An initial total of 2,302 participants with a mean age of 10.78 years (SD = 4.3) were included in the 25 CBT intervention trials incorporated in this meta-analysis. In total, the studies included 1,580 men (68.6 %) and 722 women (31.4 %). Ten studies included participants diagnosed according to ICD 9/10 or DSM IV, with CD and ODD representing the majority of diagnoses. The 25 studies examined the effect of 12 different interventions with a median treatment duration of 22 h. 1,121 participants were allocated to the CBT interventions and 1,002 participants were allocated to the control conditions. 60.0 % of the CBT interventions were group based and the majority of the studies were conducted in clinical settings (76 %). Table 1 summarizes the characteristics of the 25 studies included in this meta-analysis study. In total, 411 participants (17.8 %) dropped out of the studies after randomization, with individual study drop-out rates ranging from 3.7 to 40.0 %. The 25 trials used various approaches to assess aggression over time. The most frequently used instruments were parent-reported questionnaires (52 %). However, self-report methods were also commonly used (40 %). All studies reported pre- and posttreatment aggression scores and mean baseline data were reported for 2,046 participants (88.8 %). Furthermore, 16 studies (64 %) included a follow-up time after post-treatment, with a mean of 11.4 months and 10 studies (40 %) used blinded raters. Looking at the quality of the included studies according to the Jadad scale (5 = excellent,
123
123
6.6
6.3
3.7 (2,5–5)
15.1 (13–17)
4.1 (3–5)
13.5
6.6 (4–8)
6.6 (4–8)
9.2 (7–11)
11.2 (9–13)
15.6 (13–18)
15.1 (14–16)
15.0 (14–16)
14.5 (14–15)
15.5
4.0 (3–5)
August [44]
Brotman [46]
Butler [47]
Carpenter [48]
Down [36]
Feinfield [49]
Larsson [50]
Lipman [51]
van Manen [37]
Mitchell [52]
Nickel [54]
Nickel [56]
Nickel [55]
Nickel [53]
O’Neal [57]
60/32 (65.2/ 34.8)
–/40 (–/ 100)
72/– (100/–)
–/36 (–/ 100)
44/– (100/–)
40/– (100/–)
97/– (100/–)
102/21 (82.9/ 17.1)
101/26 (74.3/ 20.5)
40/7 (71.4/ 12.5)
16/9 (64/ 36)
43/37 (53.7/ 46.3)
89/19 (82.4/ 17.6)
19/11 (63.3/ 36.7
185/142 (56.6/ 43.4)
168/77 (68.6/ 31.4)
–
–
CD/ODD/ ADHD
–
ODD/CD/ ADHD/ BPD
Mental health
CD/ODD/ DBDNOS
–
ODD/CD
–
–
–
–
–
–
–
Reported design qualityb
Blind raters
3
5
5
Blind raters Blind raters
5
5
4
5
3
2
2
1
2
3
2
3
2
Blind raters
Blind raters
Blind raters
Blind raters
Blind raters
Open raters
Open raters
Open raters
Open raters
Open raters
Open raters
Open raters
Open raters
Blinded study?
IYP
BSFT
BSFT
FT
FT
CBT
SCIP
CBT
IYP
PT ? CT
CBT
SST
MST
IYP
ERPP
ERPP
Type CBT
Diagnoses
Age (in years)
# men/ women (%)
Intervention
Study population
August [45]
References
Table 1 Characteristics of included studies
?
20 h
20 h
16 h
24 h
?
12.8 h
24 h
64 h
27.5 h
15 h
?
?
157.5 h
245 h
334 h
Duration
47
20
36
15
22
19
42
62
55
28
18
34
56
16
111
124
Participants
Group
Family
Family
Family
Family
?
Group
Group
Group
Group
Group
Group
Family
Group
Group
Group
Type treatment
NI
PI
PI
NI
PI
TAU
WL
NI
WL
WL
PD AM
HS
TAU
NI
NI
NI
Type control
Control
–
20 h
20 h
–
?
–
–
–
–
–
15 h
?
?
–
–
–
Duration
45
20
36
15
22
21
15
61
30
28
7
46
52
14
109
121
Participants
–
Family
Family
–
Family
Individual
–
–
–
–
Group
Group
Individual
–
–
–
Type treatment
Clinic
Clinic
Clinic
Clinic
Clinic
Clinic
Clinic
Clinic
Clinic
Clinic
Clinic
School
Clinic
Clinic
School
School
Setting
35 (38.0)
4 (10.0)
9 (12.5)
5 (13.9)
7 (15.9)
16 (40.0)
?
24 (19.5)
25 (18.4)
17 (30.4)
8 (24)
9 (11.3)
4 (3.7)
10 (33.3)
78 (32.0)
44 (18.0)
Dropout n (%)a
OBS
SR
SR
SR
SR
SR
PR
PR
PR
PR
SR
PR
PR
PR
PR
OBS
Measurement
Outcome
Post, 8,16 months
Post, 12 months
Post
Post, 6 months
Post, 12 months
Post, 6, 12, 18 months
Post, 12 months
Post
Post, 12 months
Post, 5 months
Post
Post, 3 months
Post
Post, 6 months
Post, 12, 24 months
Post, 12, 24 months
Timing
0.77
2.35
0.97
0.50
1.76
-0.02
-0.13
0.25
0.71
0.84
0.76
-0.19
0.19
0.70
-0.14
0.07
ES
Eur Child Adolesc Psychiatry
Blinded study?
Reported design qualityb
80/19 (80.8/ 19.2)
68/9 (88.3/ 11.7)
14.5
15.6 (12–18)
12.3 (10–15)
9.7
15.0 (12–17)
5.9 (4–8)
10.1 (8–13)
Rowland [60]
Santiste ban [61]
Schechtman [62]
Sukhodolsky [63]
Sundell [64]
WebsterStratton [65]
Van de Wiel [66]
DBD
CD/ODD
CD
ODD/CD
–
–
SED
–
Blind raters
Open raters
Blind raters
Open raters
Open raters
Open raters
Open raters
Open raters
Open raters
4
2
5
2
1
2
2
3
2
CPP
IYP
MST
SPST
CBT
BSFT
MST
SNAP GC
MST
?
?
?
8.3 h
7.5 h
20 h
72 h
?
?
Duration
38
51
79
15
34
80
26
47
46
Participants
Group
Group
Family
Group
Group/ Indivi
Family
Family
Group
Family
Type treatment
–
–
TAUc
–
8.3 h
–
24 h
–
–
–
Duration
WL
TAU
SST
WL
PLI
TAU
WL
TAU
Type control
Control
39
48
77
16
36
46
29
40
29
Participants
Individual/ family
–
Individual/ family
Group
–
Group
–
–
–
Type treatment
Clinic
9 (11.7)
2 (2.0)
7 (4.5)
Clinic
Clinic
11 (35.5)
7 (10.0)
41 (32.5)
4 (7.3)
29 (33.3)
6 (8.0)
Dropout n (%)a
?
School
Clinic
Home
Clinic
Clinic
Setting
PR
PR
PR
SR
SR
PR
SR
PR
SR
Measurement
Outcome
Post
Post, 12 months
Post
Post, 3 months
Post
Post
Post
Post, 6 months
Post, 6, 24 months
Timing
0.38
0.35
-0.12
-0.10
0.64
0.69
0.49
0.41
0.25
ES
c
b
a
Treatment as Usual was defined differently over the studies, like ‘regular services’, ‘behavior therapy’, ‘family therapy, ‘individual counseling’ or ‘residential care’
Reported design quality of the included studies, based on the Jadad ratings, 5 = excellent, 4 = good, 3 = fair, 2 = poor, 1 = very poor
The percentage drop-out was defined as the percentage of participants not available after randomization at follow-up and was calculated using information from the studies
ODD oppositional defiant disorder, CD conduct disorder, DBD-NOS disruptive behavior disorder-not otherwise specified, ADHD attention deficit hyperactivity disorder, BPD borderline personality disorder, SED serious emotional disturbance, DBD disruptive behavior disorder, ERPP early risers prevention program, IYP incredible years program, SST social skills training, PT ? CT parent training and child training, CBT cognitive behavioral therapy, SCIP social cognitive intervention program, FT family therapy, BSFT brief strategic family therapy, MST multisystemic treatment, SNAP GC girls connected, SPST social problem-solving training, CPP coping power program, NI no intervention, HS head start, WL wait list, TAU treatment as usual, PI placebo intervention, PLI participatory learning intervention, PD AM Personal Development Anger Management intervention, OBS observer-rated, PR parent-rated, SR self-report, ES effect size, ? unknown
95/61 (60.9/ 39.1)
31/– (100/–)
55/15 (78.6/ 21.4)
95/31 (75.4/ 24.6)
32/23 (58.2/ 41.8)
–/80 (–/ 100)
8.6 (5–11)
Pepler [59]
48/27 (64.0/ 36.0)
15.1 (12–17)
–
Type CBT
Diagnoses
Age (in years)
# men/ women (%)
Intervention
Study population
Ogden [58]
References
Table 1 continued
Eur Child Adolesc Psychiatry
123
Eur Child Adolesc Psychiatry
Fig. 1 Forest plot, comparison of effect sizes per study
4 = good, 3 = fair, 2 = poor, 1 = very poor), 13 studies (52 %) were rated as ‘fair’ or higher. Effectiveness of CBT The effect sizes of the conducted studies ranged between d = -0.19 and d = 2.35 with an overall mean weighted effect size of d = 0.50, indicating a medium treatment effect for CBT to reduce aggression in adolescents. Heterogeneity was calculated (see Fig. 1) in RevMan 5.1, using the I-squared and Chi-squared test. Heterogeneity was found in the meta-analysis (I2 = 76 %, v2 = 97.99, p \ 0.0001), which shows inconsistency of study results. Therefore, the random effect model was used to correct for the heterogeneity and the different outcome measurements used in the meta-analysis. Moderators of treatment effect To explain the heterogenic outcome of the meta-analysis, we investigated whether effect sizes were moderated by specific participant and treatment characteristics. A metaregression analysis was conducted to compare the effect
123
sizes of the different studies (see Table 2). Focusing on the participant characteristics, no effect of gender and age was found regarding the treatment effectiveness of CBT. Furthermore, only 10 studies reported clinical diagnosis of the participants. The moderator analysis of these reported diagnoses did not reveal significant effects. Unfortunately, only two studies in our meta-analysis made a distinction between proactive and reactive aggression regarding the treatment of aggression (studies of [36, 37]). Because of this lack of research, proactive or reactive aggression could not be examined as moderators of treatment effect in the meta-analysis. The analysis of intervention characteristics showed that treatment duration did not significantly predict treatment effect. Also, no significant moderator effects of treatment in clinical setting, school setting or home interventions were found. Furthermore, no significant difference between the different intervention types was found (group, individual or family intervention). Of the seventeen studies that reported a first follow-up period after post-treatment, six studies reported second additional follow-up outcomes after treatment. Follow-up-treatment effect sizes ranged between d = -0.04 and d = 3.19 with an overall mean
Eur Child Adolesc Psychiatry Table 2 Values of the moderator analysis Moderators of treatment success
Level of variable (number of studies)
Gender
Boys n = 5
0.49
Girls n = 3
1.089
Boys and Girls n = 17
0.39
ES per level
Age Diagnosis n = 10
0.42 0.54
Setting
No Diagnosis n = 15 Clinic n = 19
0.61
School n = 4
0.098
Home n = 1 Unknown n = 1
F(3,24) = 1.223, p = 0.326
0.49
Group n = 15
0.35
Family n = 9
0.78
Duration of treatment
Mean = 72.94 h
Unknown n = 1
F (1,24) = 0.206, p = 0.655
-0.10
Intervention type
Jadad quality
F(2,24) = 1.916, p = 0.171
F(1,24) = 0.977, p = 0.550
Diagnosis
Rater
p value
F(2,24) = 2.058, p = 0.152
-0.02 F(12,15) = 0.354, p = 0.916
Median = 24 h \24 h n = 8
0.71
C24 h n = 8
0.58
No information n=9
0.22
Open n = 15
0.38
Blind n = 10
0.67
Very poor n = 2
0.70
Poor n = 10
0.38
Fair n = 5
0.29
Good n = 2
0.18
Excellent n = 6
0.89
F(2,24) = 1.646, p = 0.216
F(1,24) = 1.509, p = 0.232 F(4,24) = 1.124, p = 0.373
weighted effect size of d = 0.92, indicating large treatment effects on longer terms. However, follow-up effect sizes were not statistically significant different compared to posttreatment effect sizes (t = 0.42, p = 0.698). Looking at the study differences between open and blinded raters and the quality of the studies according to the Jadad scale (low versus high), no significant differences were found between effect sizes.
Discussion The aim of this study was to examine response predictors of CBT interventions in children and adolescents with aggression problems. Besides the common predictors such as gender, age, diagnosis, setting, treatment delivery and quality of the study, proactive and reactive aggressions were investigated as treatment moderators. A meta-analysis was
conducted to define these moderators. Overall, results of the meta-analysis study showed a medium treatment effect of CBT to reduce aggression in children and adolescents (d = 0.50), which is in line with earlier meta-analysis [14, 15]. The results were heterogeneous, therefore a random effect model was used and moderators of effect were investigated. Looking at participant characteristics as moderators of treatment effect in the meta-analysis, age, gender, diagnosis, type of raters and type of aggression (proactive and reactive aggression) were examined as possible predictors of response to the treatment. None of these moderators were found to predict treatment effects in the meta-analysis. Gender and age were not found as moderators of effect in a previous meta-analysis either [14, 38]. Unfortunately, proactive and reactive aggressions were only mentioned in two of the 25 included studies and therefore could not be examined as a possible predictor of response to treatment. However, the study of Down et al. [36] showed that treatment focused on reactive aggression (CBT) but also on proactive aggression (Personal Developmental Anger Management) was positive related to higher levels of anger control and self esteem after following this treatment. Age was found as a moderator of treatment effect in the reactive treatment group, where young adolescents did show lower levels of anger control and self esteem after following the treatment compared to the older adolescents. Furthermore, Van Manen et al. [37] showed that after following the Social Cognitive Intervention Program and the Social Skills Program, proactive and reactive aggression decreased in 9- to 13-year old boys after 1-year follow-up. However, participants with reactive aggression reacted significantly better after following the Social Cognitive Intervention Program (based on social information processing) than participants with proactive aggression. Since several possible moderators of effect were examined and none differed significantly from each other, the heterogeneity of the meta-analysis might be caused by the lack of differentiation between proactive and reactive aggression in the majority of the studies included in this meta-analyses. Overall, the meta-analysis indicates that the external and in particular predictive validity of proactive and reactive aggression regarding the response to CBT needs to be further investigated. The diagnoses of the participants were only reported in 10 of the 25 studies. No significant effect was found in these 10 studies that reported the diagnosis of the participants (overall ODD/ CD), however this could be due to the fact of lacking data. In future research, studies should report their samples in more detail regarding behavioral symptoms and clinical diagnosis of the participants as this could be a possible moderator variable. For example, it has been shown that reactive aggression is associated with anxiety and attention
123
Eur Child Adolesc Psychiatry
problems and proactive aggression is associated with delinquent behavior and CD [18]. More insight in these possible moderator variables, could lead to more effective and personalized interventions. Finally, treatment characteristics were examined as possible moderators of treatment response. Results indicated that treatment duration, treatment setting (school versus clinical) and intervention type (e.g., individual, group or family) did not predict treatment outcome of CBT. The finding that treatment duration does not seem to influence outcome is quite remarkable, but is in line with results of previous meta-analyses [14, 39]. This may suggest that more cost-effective development of future interventions is feasible. Moreover, there is even some evidence suggesting that recidivism is lower after treatment of shorter duration versus longer duration [40, 41]. This could possibly be explained by that short interventions create a larger momentum and/or are more intensive. The finding that treatment setting also does not seem to influence outcome is hopeful, since it means that less-invasive (and often cost-effective) school-based interventions may be preferred compared to clinical interventions. However, there are studies suggesting that clinical interventions are more effective than school-based interventions in clinical severe cases, suggesting it is premature to prefer schoolbased interventions over clinical interventions in all cases [39]. Lastly, intervention type did not seem to moderate treatment outcome, although a meta-analysis did indicate that including parents in the treatment of adolescents is important [15]. Since the lack of power could have caused the non-significant results of the moderator analysis, future research should examine these possible moderators of treatment effect in larger meta-analyses to develop more effective and cost-beneficial treatment of aggression. Also, the meta-analysis did not find significant difference between effect sizes of studies with open raters or blinded studies, or between that of studies with high versus low quality. However, this could be due to that different outcome measurements were used in the studies included in our meta-analysis. This could lead to a bias in the results, therefore studies with matching outcome measurements are highly needed in future to have a better indication of the overall outcomes regarding the treatment of aggression. Limitations This study had some limitations. First of all, a possible limitation of the present meta-analysis is that the search strategy primarily relied on published studies. As a result, magnitudes of treatment effects might have been overestimated [42]. On the other hand, it has been argued that including grey literature could also lead to a bias of the results, since grey literature often includes non-peer
123
reviewed articles with lower level of methodologically quality [43]. Furthermore, only a small number of studies were included in the moderator analyses, which could have biased the results of the moderator analyses due to a lack of power.
Conclusion The findings of this meta-analysis indicate that CBT is effective in decreasing aggression in adolescents (d = 0.50). Treatment style, age and gender were not found as moderators of response to treatment. Too few studies examined the moderating role of proactive and reactive aggression as predictors of response to treatment, and thus further research is needed in this area. Also, many studies did insufficiently report on clinical diagnosis and behavioral symptoms of the study sample. Furthermore, it was shown that treatment setting and duration did not seem to influence treatment effect, which shows the need for development of more cost-effective and lessinvasive interventions. Finally, better standardization of design, predictors, and outcome measures across studies is required to establish predictive profiles of responders and non-responders to treatment of reducing aggression. Conflict of interest Jan K. Buitelaar has been a consultant to/ member of advisory board of/and/or speaker for Janssen Cilag BV, Eli Lilly, Bristol-Myer Squibb, Shering Plough, UCB, Shire, Novartis and Servier. He is not an employee of any of these companies, and not a stock shareholder of any of these companies. He has no other financial or material support, including expert testimony, patents, royalties. F.E.Scheepers has travel support and is a speaker for Eli Lilly. K. Smeets, A.Leeijen, M. van der Molen and N. Rommelse state no conflict of interest.
References 1. Armbruster MA, Sukhodolsky DG, Michalsen R (2004) The impact of managed care on children’s outpatient treatment: a comparison study of treatment outcome before and after managed care. Am J Orthopsychiatry 74(1):5–13 2. Rutter M, Bishop D, Pine D, Scott S, Stevenson JS, Taylor EA, Thapar A (2010) Rutter’s child and adolescent psychiatry. Wiley, Oxford 3. Gannon TA, Ward T, Beech AR, Fisher D (2007) Aggressive offenders’ cognition: theory, research and practice. Wiley, Chichester 4. Romeo R, Knapp M, Scott S (2006) Economic cost of severe antisocial behaviour in children—and who pays it. Br J Psychiatry 18(8):547–553 5. Buitelaar JK, Smeets KC, Herpers P, Scheepers F, Glennon J, Rommelse NNJ (2013) Conduct disorders. Eur Child Adolesc Psychiatry 22(1):49–54 6. Frick PJ, Viding E (2009) Antisocial behavior from a developmental psychopathology perspective. Dev Psychopathol 21(04): 1111–1131
Eur Child Adolesc Psychiatry 7. Monahan KC, Steinberg L, Cauffman E (2009) Affiliation with antisocial peers, susceptibility to peer influence, and antisocial behavior during the transition to adulthood. Dev Psychol 45(6):1520–1530 8. Armelius BA, Andreassen TH (2007) Cognitive-behavioral treatment for antisocial behavior in youth in residental treatment. Cochrane Database Syst Rev 4:1–35 9. Connor DF, Carlson GA, Chang KD, Daniolos PT, Ferziger R, Findling RL, Hutchinson JG, Malone RP, Halperin JM, Plattner B, Post RM, Reynolds DL, Rogers KM, Saxena K, Steiner H (2006) Juvenile maladaptive aggression: a review of prevention, treatment, and service configuration and a proposed research agenda. J Clin Psychiatry 67(5):808–820 10. Barzman DH, Findling RL (2008) Pharmacological treatment of pathologic aggression in children. Int Rev Psychiatry 20:151–157 11. Mattaini MA, McGuire MS (2006) Behavioral strategies for constructing nonviolent cultures with youth: a review. Behav Modif 30(2):184–224 12. Blake CS, Hamrin V (2007) Current approaches to the assessment and management of anger and aggression in youth: a review. J Child Adolesc Psychiatr Nurs 20(4):209–221 13. Aman MG, Bukstein OG, Gadow KD, LE Arnold, Molina BS, McNamara NK, Rundberg-Rivera EV, Li X, Kipp H, Schneider J, Butter EM, Baker J, Sprafkin J, Rice RR Jr, Bangalore SS, Farmer CA, Austin AB, Buchan-Page KA, Brown NV, Hurt EA, Grondhuis SN, Findling RL (2014) What does risperidone add to parent training and stimulant for severe aggression in child attention-deficit/hyperactivity disorder? J Am Acad Child Adolesc Psychiatry 53(1):47–60. doi:10.1016/j.jaac.2013.09.022 e41 14. Sukhodolsky DG, Kassinove H, Gorman BS (2004) Cognitivebehavioral therapy for anger in children and adolescents: a metaanalysis. Aggress Viol Behav 9(3):247–269 15. Fossum S, Handega˚rd BH, Martinussen M, Mørch WT (2008) Psychosocial interventions for disruptive and aggressive behaviour in children and adolescents: a meta-analysis. Eur Child Adolesc Psychiatry 17:438–451 16. Sonuga-Barke EJ, Brandeis D, Cortese S, Daley D, Ferrin M, Holtmann M, Stevenson J, Danckaerts M, van der Oord S, Dopfner M, Dittmann RW, Simonoff E, Zuddas A, Banaschewski T, Buitelaar J, Coghill D, Hollis C, Konofal E, Lecendreux M, Wong IC, Sergeant J (2013) Nonpharmacological interventions for ADHD: systematic review and meta-analyses of randomized controlled trials of dietary and psychological treatments. Am J Psychiatry 170(3):275–289 17. Polman H, Orobio de Castro B, Koops W, van Boxtel HW, Merk WM (2007) A meta-analysis of the distinction between reactive and pro active aggression in children and adolescents. J Abnorm Child Psychol 35:522–535 18. Vitaro F, Brendgen M, Tremblay RE (2002) Reactively and proactively aggressive children: antecedent and subsequent characteristics. J Child Psychol Psychiatry 43(4):495–505 19. Stanford MS, Houston RJ, Mathias CW, Villemarette-Pittman NR, Helfritz LE, Conklin SM (2003) Characterizing aggressive behavior. Assessment 10(2):183–190 20. Kockler TR, Stanford MS, Nelson CE, Meloy JR, Sanford K (2006) Characterizing Aggressive Behavior in a Forensic Population. Am J Orthopsychiatry 76(1):80–85 21. Frick PJ, Ellis M (1999) Callous-unemotional traits and subtypes of conduct disorder. Clin Child Fam Psychol Rev 2(3):149–168 22. Blair RJ (2001) Neurocognitive models of aggression, the antisocial personality disorders, and psychopathy. J Neurol Neurosurg Psychiatry 71(6):727–731 23. Blair RJ, Peschardt KS, Budhani S, Mitchell DG, Pine DS (2006) The development of psychopathy. J Child Psychol Psychiatry 47(3–4):262–276
24. Vitaro F, Brendgen M, Barker ED (2006) Subtypes of aggressive behaviors: a developmental perspective. Int J Behav Dev 30(1):12–19 25. Barker ED, Vitaro F, Lacourse E, Fontaine NMG, Carbonneau R, Tremblay RE (2010) Testing the developmental distinctiveness of male proactive and reactive aggression with a nested longitudinal experimental intervention. Aggress Behav 36:127–140 26. Haas SM, Waschbusch DA, Pelham WE Jr, King S, Andrade BF, Carrey NJ (2011) Treatment response in CP/ADHD children with callous/unemotional traits. J Abnorm Child Psychol 39:541–552 27. Masi G, Manfredi A, Milone A, Muatori P, Polidori L, Ruglioni L, Muratori F (2011) Predictors of nonresponse to psychosocial treatment in children and adolescents with disruptive behaviour disorders. J Child Adolesc Psychopharmacol 21(1):51–55 28. Card NA, Little TD (2006) Proactive and reactive aggression in childhood and adolescence: a meta-analysis of differential relations with psychosocial adjustment. Int J Behav Dev 30(5): 466–480. doi:10.1177/0165025406071904 29. Poulin F, Boivin M (2000) The role of proactive and reactive aggression in the formation and development of boys’ friendships. Dev Psychol 36(2):233–240 30. Bushman BJ, Anderson CA (2001) Is it time to pull the plug on the hostile versus instrumental aggression dichotomy? Psychol Rev 108(1):273–279 31. Morris SB (2008) Estimating effect sizes from pretest-posttestcontrol group designs. Organ Res Methods 11(2):364–386 32. Lipsey MW, Landenberger NA, Wilson SJ (2007) Effects of cognitive-behavioral programs for criminal offenders. Campbell Syst Rev 6. doi:10.4073/csr.2007.6 33. Jadad AR, Moore RA, Carroll D, Jenkinson C, Reynolds DJ, Gavaghan DJ, McQuay HJ (1996) Assessing the quality of reports of randomized clinical trials: is blinding necessary? Control Clin Trials 17(1):1–12 34. Cohen J (1988) Statistical power analysis for the behavioral sciences. Lawrence Erlbaum Associations, Hillsdale 35. Nordic Cochrane Centre TCC (2011) Review Manager (RevMan). (Version 5.1) 36. Down R, Willner P, Watts L, Griffiths J (2011) Anger Management groups for adolescents: a mixed-methods study of efficacy and treatment preferences. Clin Child Psychol Psychiatry 16(1):33–52 37. Van Manen TG, Prins PJ, Emmelkamp PM (2004) Reducing aggressive behavior in boys with a social cognitive group treatment: results of a randomized, controlled trial. J Am Acad Child Adolesc Psychiatry 43(12):1478–1487 38. Landenberger NA, Lipsey MW (2005) The positive effects of cognitive–behavioral programs for offenders: a meta-analysis of factors associated with effective treatment. J Exp Criminol 1:451–476 39. McCart MR, Priester PE, Davies WH, Azen R (2006) Differential effectiveness of behavioral parent-training and cognitive-behavioral therapy for antisocial youth: a meta-analysis. J Abnorm Child Psychol 34(4):527–543 40. Dekovic M, Slagt MI, Asscher JJ, Boendermaker L, Eichelsheim VI, Prinzie P (2011) Effects of early prevention programs on adult criminal offending: a meta-analysis. Clin Psychol Rev 31(4):532–544. doi:10.1016/j.cpr.2010.12.003 41. van der Put CE, Asscher JJ, Stams GJ, van der Laan PH, Breuk R, Jongman E, Doreleijers T (2013) Recidivism after treatment in a forensic youth-psychiatric setting: the effect of treatment characteristics. Int J Offender Ther Comp Criminol 57(9):1120–1139. doi:10.1177/0306624x12452389 42. Dickersin K (1990) The existence of publication bias and risk factors for its occurrence. J Am Med Assoc 263(10):1385–1389
123
Eur Child Adolesc Psychiatry 43. Hopewell S, McDonald S, Clarke MJ, Egger M (2007) Grey literature in meta-analyses of randomized trials of health care interventions. Cochrane Database Syst Rev 2:1–16 55.
Studies included in the meta-analysis 44. August GJ, Lee SS, Bloomquist ML, Realmuto GM, ektner JM (2003) Dissemination of an evidence-based prevention innovation for aggressive children living in culturally diverse, urban neighborhoods: the Early Risers effectiveness study. Prev Sci 4(4):271–286 45. August GJ, Realmuto GM, Hektner JM, Bloomquist ML (2001) An integrated components preventive intervention for aggressive elementary school children: the early risers program. J Consult Clin Psychol 69(4):614–626 46. Brotman LM, Klein RG, Kamboukos D, Brown EJ, Coard SI, Sosinsky LS (2003) Preventive intervention for urban, lowincome preschoolers at familial risk for conduct problems: a randomized pilot study. J Clin Child Adolesc Psychol 32(2):246–257 47. Butler S, Baruch G, Hickey N (2011) A randomized controlled trial of multisystemic therapy and a statutory therapeutic intervention for young offenders. J Am Acad Child Adolesc Psychiatry 50(1220–1235):e1222 48. Carpenter EM (2002) A curriculum-based approach for socialcognitive skills training: an intervention targeting aggression in Head Start preschoolers. University of Maine 49. Feinfield KA, Baker BL (2004) Empirical support for a treatment program for families of young children with externalizing problems. J Clin Child Adolesc Psychol 33(1):182–195 50. Larsson B, Fossum S, Clifford G, Drugli MB, Handega˚rd BH, Mørch WT (2009) Treatment of oppositional defiant and conduct problems in young Norwegian children : results of a randomized controlled trial. Eur Child Adolesc Psychiatry 18(1):42–52 51. Lipman EL, Boyle MH, Cunningham C, Kenny M, Sniderman C, Duku E, Mills B, Evans P, Waymouth M (2006) Testing effectiveness of a community-based aggression management program for children 7–11 years old and their families. J Am Acad Child Adolesc Psychiatry 45(9):1085–1093 52. Mitchell P, Smedley K, Kenning C, McKee A, Woods D, Rennie CE, Bell RV, Aryamanesh M, Dolan M (2011) Cognitive behaviour therapy for adolescent offenders with mental health problems in custody. J Adolesc Health 34(3):433–443 53. Nickel M, Luley J, Krawczyk J, Nickel C, Widermann C, Lahmann C, Muehlbacher M, Forthuber P, Kettler C, Leiberich P, Tritt K, Mitterlehner F, Kaplan P, Pedrosa Gil F, Rother W, Loew T (2006) Bullying girls—changes after brief strategic family therapy: a randomized, prospective, controlled trial with 1-year follow-up. Psychother Psychosom 75(1):47–55 54. Nickel MK, Krawczyk J, Nickel C, Forthuber P, Kettler C, Leiberich P, Muehlbacher M, Tritt K, Mitterlehner FO, Lahmann C, Rother WK, Loew TH (2005) Anger, interpersonal
123
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
relationships, and health-related quality of life in bullying boys who are treated with outpatient family therapy: a randomized, prospective, controlled trial with 1 year of follow-up. Pediatrics 116(2):247–254 Nickel MK, Muehlbacher M, Kaplan P, Krawczyk J, Buschmann W, Kettler C, Rother N, Egger C, Rother WK, Loew TK, Nickel C (2006) Influence of family therapy on bullying behaviour, cortisol secretion, anger, and quality of life in bullying male adolescents: a randomized, prospective, controlled study. Can J Psychiatry 51(6):355–362 Nickel MK, Nickel C, Leiberick R, Tritt K, Mitterlehner FO, Lahmann C, Rother WK, Loew TH (2005) Aggressive female youth benefit from outpatient family therapy: a randomized, prospective, controlled trial. Pediatri Int 47(2):167–171 O’Neal CR, Brotman LM, Huang KY, Gouley KK, Kamboukos D, Calzada EJ, Pine DS (2010) Understanding relations among early family environment, cortisol response, and child aggression via a prevention experiment. Child Dev 81(1):290–305 Ogden T, Hagen KA (2006) Multisystemic treatment of serious behaviour problems in youth: sustainability of effectiveness 2 years after intake. Child Adolesc Mental Health 11(3):142–149 Pepler D, Walsh M, Yuile A, Levene K, Jiang D, Vaughan A, Webber J (2010) Bridging the gender gap: interventions with aggressive girls and their parents. Prev Sci 11(3):229–238 Rowland MD, Halliday-Boykins CA, Henggeler SW, Cunningham PB, Lee TG, Kruesi MJP, Shapiro SB (2005) A randomized trial of multisystemic therapy with Hawaii’s felix class youths. J Emot Behav Disord 13(1):13–23 Santisteban DA, Coatsworth JD, Perez-Vidal A, Kurtines WM, Schwartz SJ, LaPerriere A, Szapocznik J (2003) Efficacy of brief strategic family therapy in modifying Hispanic adolescent behavior problems and substance use. J Fam Psychol 17(1): 121–133 Schechtman Z (2000) An innovative intervention for treatment of children and adolescent aggression: an outcome study. Psychol Schools 37(2):157–167 Sukhodolsky DG, Golub A, Stone EC, Orban L (2005) Dismantling anger control training for children: a randomized pilot study of social problem-solving versus social skills training components. Behav Ther 36:15–23 Sundell K, Hansson K, Lo¨fholm CA, Olsson T, Gustle LH, Kadesjo¨ C (2008) The transportability of multisystemic therapy to Sweden: short-term results from a randomized trial of conductdisordered youths. J Fam Psychol 22(4):550–560 Webster-Stratton C, Reid J, Hammond M (2001) Social skills and problem-solving training for children with early-onset conduct problems: who benefits? J Child Psychol Psychiatry 42(7): 943–952 van de Wiel NM, Matthys W, Cohen-Kettenis PT, Maassen GH, Lochman JE, van Engeland H (2007) The effectiveness of an experimental treatment when compared to care as usual depends on the type of care as usual. Behav Modif 31(3):298–312
REVIEW
Treatment moderators of cognitive behavior therapy to reduce aggressive behavior: a meta-analysis Kirsten C. Smeets • Anouk A. M. Leeijen • Marie¨t J. van der Molen • Floor E. Scheepers Jan K. Buitelaar • Nanda N. J. Rommelse
•
Received: 5 November 2013 / Accepted: 25 July 2014 Ó Springer-Verlag Berlin Heidelberg 2014
Abstract Maladaptive aggression in adolescents is an increasing public health concern. Cognitive Behavior Therapy (CBT) is one of the most common and promising treatments of aggression. However, there is a lack of information on predictors of treatment response regarding CBT. Therefore, a meta-analysis was performed examining the role of predictors on treatment response of CBT. Twenty-five studies were evaluated (including 2,302 participants; 1,580 boys and 722 girls), and retrieved through searches on PubMed, PsycINFO and EMBASE. Effect sizes were calculated for studies that met inclusion criteria. Study population differences and specific CBT characteristics were examined for their explanatory power. There was substantial variation across studies in design and outcome variables. The meta-analysis showed a medium treatment effect for CBT to reduce aggression (Cohen’d = 0.50). No predictors of treatment response were found in the meta-analysis. Only two studies did Trial registration: ISRCTN21681959.
Electronic supplementary material The online version of this article (doi:10.1007/s00787-014-0592-1) contains supplementary material, which is available to authorized users. K. C. Smeets (&) J. K. Buitelaar N. N. J. Rommelse Karakter Child and Adolescent Psychiatry University Centre, Reinier Postlaan 12, 6525 GC Nijmegen, The Netherlands e-mail: [email protected] A. A. M. Leeijen Department of Health Sciences, VU University Amsterdam, Amsterdam, The Netherlands M. J. van der Molen Faculty of Psychology and Education and the EMGO Institute for Health and Care Research, VU University Amsterdam, Amsterdam, The Netherlands
examine whether proactive versus reactive aggression could be a moderator of treatment outcome, and no effect was found of this subtyping of aggression. These study results suggest that CBT is effective in reducing maladaptive aggression. Furthermore, treatment setting and duration did not seem to influence treatment effect, which shows the need for development of more cost-effective and less-invasive interventions. More research is needed on moderators of outcome of CBT, including proactive versus reactive aggression. This requires better standardization of design, predictors, and outcome measures across studies. Keywords Aggression Cognitive behavior therapy Meta-analysis Effect size
Introduction Maladaptive aggressive behavior is common in adolescents and one of the most prevalent reasons for referral to mental health services [1, 2]. Aggression can be defined as behavior directed at an object, human or animal, through which the object, human or animal experiences harm or F. E. Scheepers Department of Psychiatry, Rudolf Magnus Institute of Neuroscience, UMC Utrecht, Utrecht, The Netherlands J. K. Buitelaar Department of Cognitive Neuroscience, Donders Institute for Brain, Cognition and Behavior, Radboud University Medical Center, Nijmegen, The Netherlands N. N. J. Rommelse Department of Psychiatry, Radboud University Medical Center, Nijmegen, The Netherlands
123
Eur Child Adolesc Psychiatry
damage [3]. Severe maladaptive aggression in adolescents is accompanied by substantial community costs [4]. One of the reasons is that aggression often demands immediate attention, because aggressive behavior leads to a variety of problems in different life domains, like being expelled from school or contact with police or justice. Children and adolescents with symptoms of aggressive behavior and disruptive behavior problems are often diagnosed with Oppositional Defiant Disorder (ODD) or Conduct Disorder (CD) [5]. Also, aggression in childhood or adolescents persists often in severe aggression in adulthood [6, 7]. A literature review shows that 86 % of the children with CD at the age of seven still have conduct problems at the age of fifteen. Furthermore, 5–10 % of all children with antisocial behavior problems continue to present with serious problems in adulthood, and 50–60 % of the crimes in youth are committed by this persistent group [8]. To reduce these costs and public health concerns, effective treatment of aggression is needed. Both pharmacologic and non-pharmacologic approaches have been applied to treat aggression in adolescents [9–13]. Atypical antipsychotics, of which Risperidone is the most commonly studied, presynaptic alpha-adrenergic agonists and psychostimulants have shown to be effective in reducing aggression in adolescents [9, 10, 13]. Furthermore, Cognitive Behavioral Therapies (CBT) such as anger management, multimodal interventions and skills training have shown to be effective in reducing aggression [14]. Several treatment guidelines recommend psychosocial treatment as the first option of treatment of aggression in children and adolescents, with consideration of pharmacologic treatment in patients who do not respond (no improvement of their aggressive behavior) to the psychosocial treatment [5]. However, there is a lack of information on predictors of treatment success regarding CBT [14, 15]. More specific knowledge of the characteristics of participants that respond to interventions could help clinicians identify the best treatment of aggression for individual patients. Characteristics of the participants (e.g., age, gender), treatment (e.g., duration, format or setting) or quality of the study design (e.g., open versus blinded raters) can be considered as possible moderators of treatment effect [16]. For example, CBT in older children seems to be more effective than CBT in younger children. Also, the number of sessions (around 20 sessions) is positively associated with the outcome of the treatment [14]. In addition, studies reporting clinical diagnoses of the participants have larger effect sizes than studies that do not report diagnoses [15]. Little attention has been paid so far to the distinction between reactive and proactive aggression as moderators of treatment effect. There is a body of evidence on the distinction between the two types of aggression, and treatment regarding proactive and reactive
123
aggression might be different [17, 18]. Reactive aggression is also commonly known as ‘impulsive’, ‘hot-blooded’, ‘affective’ or ‘defensive’ aggression and refers to emotionally charged responses to provocations or frustration [19, 20]. Proactive aggression is also commonly called ‘instrumental’, ‘cold-blooded’ or ‘predatory’ aggression and refers to conscious and planned acts that are often enacted on the basis of egocentric motives for personal gains [21–23]. The distinction between these types of aggression may be relevant in predicting treatment success. Reactive aggressors seem to benefit more from psychosocial interventions than proactive aggressors [24, 25]. Children and adolescents with severe problems of CD and proactive aggression have found to be more resistant for treatment and often do not clinically improve after following psychosocial treatment [25–27]. On the other hand, proactive and reactive aggression are often highly correlated [28, 29], and there is ongoing debate whether proactive and reactive aggression need to be distinguished or are components of one overarching construct [30]. Therefore, a meta-analysis was conducted of CBT studies that aimed to reduce aggression in children and adolescents, including their predictors of treatment outcome. Besides the common predictors such as gender, age, diagnosis, setting or treatment format, proactive and reactive aggression were also investigated as potential predictors of success regarding CBT.
Methods A systematic and comprehensive search for studies in English for the period January 2000 until April 2013 was conducted. Since previous meta-analyses regarding CBT of aggression in children and adolescents were based on literature of the period between 1968 and 2005 [14, 15], this meta-analysis was based on the most recent literature. PubMed, PsycINFO and EMBASE were searched for cognitive behavioral therapies used to treat maladaptive aggression. The search included, among others, the following keywords: aggression, cognitive behavioral therapy and randomized controlled trial (see Supplementary material Appendix I for the complete list of search terms). In addition to searching these databases, reference lists from all retrieved articles and relevant published reviews (i.e., [14, 15]) were hand-searched to identify additional related publications (see Supplementary material Appendix II for a flow chart of the search strategy). Within each individual study, outcome measures relevant to maladaptive aggression were selected during the coding stage. Only one outcome measurement per study was selected, to secure independency of the data. If more outcome measurements were present in the study, the most commonly
Eur Child Adolesc Psychiatry
used (i.e., CBCL questionnaire) measurement was chosen. Next, the effect sizes of the treatment were calculated; see below for details [31]. Studies were included if: 1.
2.
3.
4.
Cognitive behavioral strategies had been used such as coaching and modeling, and if participants’ behavior and thinking styles were addressed using techniques such as anger management, social skills training or assertiveness training [15, 32]. An experimental or quasi-experimental design had been used that assigned participants to either CBT intervention or control group (thus only Randomized Controlled Trials were included). Acceptable comparison conditions were placebo, waiting list, or treatment as usual conditions. They reported at least one quantitative measure on aggressive behavior (e.g., rating scale or observation scale, pre- and post-measurements) subsequent to treatment as an outcome variable. The CBT treatment included participants under the age of 23 with maladaptive aggression problems.
Studies were excluded if no intervention or control treatment was defined. Studies where part of the participants received medication treatment were not excluded. The following possible moderators of treatment outcome were extracted and analyzed: age, gender, diagnosis of ODD/CD, duration of treatment, setting of treatment and type of CBT treatment (family, group or individual). Furthermore, we coded whether the study outcome was assessed by non-blinded raters who were aware of treatment allocation, or by blinded raters. The quality of the study was evaluated based on the Jadad scale (rating regarding quality of randomization, blinding and treatment of missing data) [33]. These variables regarding study quality were also taken into account as possible moderators of treatment effect [16]. Statistical analyses Within each individual study, outcome measures relevant to maladaptive aggression were selected during the coding stage. First, the studies were selected by one of the authors. In cases of doubt and after the first selection was made, another author coded the studies as well (double-coding). The pretest–posttest-control group design was used to analyze the treatment effects. Effect sizes were calculated using the dppc2 developed by Morris [31]: dppc2 ¼ Cp
ðMPost;T MPre;T Þ ðMPost;C MPre;C Þ SDPre
where Mpost, T referred to the mean treatment population aggression score post test, Mpre, T referred to the mean treatment population aggression score pre-test, Mpost, C
referred to the mean control population aggression score post test and Mpre, C referred to the mean control population aggression score pre-test. SDpre, referred to the standard deviation of the pre-test and Cp is the sample size bias correction, based on the n in the treatment group (Nt) and the n in the control group (Nc) [31]: Cp¼1
3 1 4ðNt þ Nc 2Þ
An effect size of d = 0.20 indicates a small effect, an effect size of d = 0.50 denotes a medium effect and an effect size of d = 0.80 represents a large effect. Furthermore, negative effect sizes assume favorable effects in control groups [34]. Also, heterogeneity was calculated in Review Manager 5.1 (RevMan) [35]. A meta-regression analysis was conducted to determine the different possible moderators of effect (sensitivity analysis). The effect of moderators on the treatment effect sizes was analyzed by a t test or linear regression analysis for continuous variables and Chi-square test for categorical variables, using SPSS 20.0.
Results An initial total of 2,302 participants with a mean age of 10.78 years (SD = 4.3) were included in the 25 CBT intervention trials incorporated in this meta-analysis. In total, the studies included 1,580 men (68.6 %) and 722 women (31.4 %). Ten studies included participants diagnosed according to ICD 9/10 or DSM IV, with CD and ODD representing the majority of diagnoses. The 25 studies examined the effect of 12 different interventions with a median treatment duration of 22 h. 1,121 participants were allocated to the CBT interventions and 1,002 participants were allocated to the control conditions. 60.0 % of the CBT interventions were group based and the majority of the studies were conducted in clinical settings (76 %). Table 1 summarizes the characteristics of the 25 studies included in this meta-analysis study. In total, 411 participants (17.8 %) dropped out of the studies after randomization, with individual study drop-out rates ranging from 3.7 to 40.0 %. The 25 trials used various approaches to assess aggression over time. The most frequently used instruments were parent-reported questionnaires (52 %). However, self-report methods were also commonly used (40 %). All studies reported pre- and posttreatment aggression scores and mean baseline data were reported for 2,046 participants (88.8 %). Furthermore, 16 studies (64 %) included a follow-up time after post-treatment, with a mean of 11.4 months and 10 studies (40 %) used blinded raters. Looking at the quality of the included studies according to the Jadad scale (5 = excellent,
123
123
6.6
6.3
3.7 (2,5–5)
15.1 (13–17)
4.1 (3–5)
13.5
6.6 (4–8)
6.6 (4–8)
9.2 (7–11)
11.2 (9–13)
15.6 (13–18)
15.1 (14–16)
15.0 (14–16)
14.5 (14–15)
15.5
4.0 (3–5)
August [44]
Brotman [46]
Butler [47]
Carpenter [48]
Down [36]
Feinfield [49]
Larsson [50]
Lipman [51]
van Manen [37]
Mitchell [52]
Nickel [54]
Nickel [56]
Nickel [55]
Nickel [53]
O’Neal [57]
60/32 (65.2/ 34.8)
–/40 (–/ 100)
72/– (100/–)
–/36 (–/ 100)
44/– (100/–)
40/– (100/–)
97/– (100/–)
102/21 (82.9/ 17.1)
101/26 (74.3/ 20.5)
40/7 (71.4/ 12.5)
16/9 (64/ 36)
43/37 (53.7/ 46.3)
89/19 (82.4/ 17.6)
19/11 (63.3/ 36.7
185/142 (56.6/ 43.4)
168/77 (68.6/ 31.4)
–
–
CD/ODD/ ADHD
–
ODD/CD/ ADHD/ BPD
Mental health
CD/ODD/ DBDNOS
–
ODD/CD
–
–
–
–
–
–
–
Reported design qualityb
Blind raters
3
5
5
Blind raters Blind raters
5
5
4
5
3
2
2
1
2
3
2
3
2
Blind raters
Blind raters
Blind raters
Blind raters
Blind raters
Open raters
Open raters
Open raters
Open raters
Open raters
Open raters
Open raters
Open raters
Blinded study?
IYP
BSFT
BSFT
FT
FT
CBT
SCIP
CBT
IYP
PT ? CT
CBT
SST
MST
IYP
ERPP
ERPP
Type CBT
Diagnoses
Age (in years)
# men/ women (%)
Intervention
Study population
August [45]
References
Table 1 Characteristics of included studies
?
20 h
20 h
16 h
24 h
?
12.8 h
24 h
64 h
27.5 h
15 h
?
?
157.5 h
245 h
334 h
Duration
47
20
36
15
22
19
42
62
55
28
18
34
56
16
111
124
Participants
Group
Family
Family
Family
Family
?
Group
Group
Group
Group
Group
Group
Family
Group
Group
Group
Type treatment
NI
PI
PI
NI
PI
TAU
WL
NI
WL
WL
PD AM
HS
TAU
NI
NI
NI
Type control
Control
–
20 h
20 h
–
?
–
–
–
–
–
15 h
?
?
–
–
–
Duration
45
20
36
15
22
21
15
61
30
28
7
46
52
14
109
121
Participants
–
Family
Family
–
Family
Individual
–
–
–
–
Group
Group
Individual
–
–
–
Type treatment
Clinic
Clinic
Clinic
Clinic
Clinic
Clinic
Clinic
Clinic
Clinic
Clinic
Clinic
School
Clinic
Clinic
School
School
Setting
35 (38.0)
4 (10.0)
9 (12.5)
5 (13.9)
7 (15.9)
16 (40.0)
?
24 (19.5)
25 (18.4)
17 (30.4)
8 (24)
9 (11.3)
4 (3.7)
10 (33.3)
78 (32.0)
44 (18.0)
Dropout n (%)a
OBS
SR
SR
SR
SR
SR
PR
PR
PR
PR
SR
PR
PR
PR
PR
OBS
Measurement
Outcome
Post, 8,16 months
Post, 12 months
Post
Post, 6 months
Post, 12 months
Post, 6, 12, 18 months
Post, 12 months
Post
Post, 12 months
Post, 5 months
Post
Post, 3 months
Post
Post, 6 months
Post, 12, 24 months
Post, 12, 24 months
Timing
0.77
2.35
0.97
0.50
1.76
-0.02
-0.13
0.25
0.71
0.84
0.76
-0.19
0.19
0.70
-0.14
0.07
ES
Eur Child Adolesc Psychiatry
Blinded study?
Reported design qualityb
80/19 (80.8/ 19.2)
68/9 (88.3/ 11.7)
14.5
15.6 (12–18)
12.3 (10–15)
9.7
15.0 (12–17)
5.9 (4–8)
10.1 (8–13)
Rowland [60]
Santiste ban [61]
Schechtman [62]
Sukhodolsky [63]
Sundell [64]
WebsterStratton [65]
Van de Wiel [66]
DBD
CD/ODD
CD
ODD/CD
–
–
SED
–
Blind raters
Open raters
Blind raters
Open raters
Open raters
Open raters
Open raters
Open raters
Open raters
4
2
5
2
1
2
2
3
2
CPP
IYP
MST
SPST
CBT
BSFT
MST
SNAP GC
MST
?
?
?
8.3 h
7.5 h
20 h
72 h
?
?
Duration
38
51
79
15
34
80
26
47
46
Participants
Group
Group
Family
Group
Group/ Indivi
Family
Family
Group
Family
Type treatment
–
–
TAUc
–
8.3 h
–
24 h
–
–
–
Duration
WL
TAU
SST
WL
PLI
TAU
WL
TAU
Type control
Control
39
48
77
16
36
46
29
40
29
Participants
Individual/ family
–
Individual/ family
Group
–
Group
–
–
–
Type treatment
Clinic
9 (11.7)
2 (2.0)
7 (4.5)
Clinic
Clinic
11 (35.5)
7 (10.0)
41 (32.5)
4 (7.3)
29 (33.3)
6 (8.0)
Dropout n (%)a
?
School
Clinic
Home
Clinic
Clinic
Setting
PR
PR
PR
SR
SR
PR
SR
PR
SR
Measurement
Outcome
Post
Post, 12 months
Post
Post, 3 months
Post
Post
Post
Post, 6 months
Post, 6, 24 months
Timing
0.38
0.35
-0.12
-0.10
0.64
0.69
0.49
0.41
0.25
ES
c
b
a
Treatment as Usual was defined differently over the studies, like ‘regular services’, ‘behavior therapy’, ‘family therapy, ‘individual counseling’ or ‘residential care’
Reported design quality of the included studies, based on the Jadad ratings, 5 = excellent, 4 = good, 3 = fair, 2 = poor, 1 = very poor
The percentage drop-out was defined as the percentage of participants not available after randomization at follow-up and was calculated using information from the studies
ODD oppositional defiant disorder, CD conduct disorder, DBD-NOS disruptive behavior disorder-not otherwise specified, ADHD attention deficit hyperactivity disorder, BPD borderline personality disorder, SED serious emotional disturbance, DBD disruptive behavior disorder, ERPP early risers prevention program, IYP incredible years program, SST social skills training, PT ? CT parent training and child training, CBT cognitive behavioral therapy, SCIP social cognitive intervention program, FT family therapy, BSFT brief strategic family therapy, MST multisystemic treatment, SNAP GC girls connected, SPST social problem-solving training, CPP coping power program, NI no intervention, HS head start, WL wait list, TAU treatment as usual, PI placebo intervention, PLI participatory learning intervention, PD AM Personal Development Anger Management intervention, OBS observer-rated, PR parent-rated, SR self-report, ES effect size, ? unknown
95/61 (60.9/ 39.1)
31/– (100/–)
55/15 (78.6/ 21.4)
95/31 (75.4/ 24.6)
32/23 (58.2/ 41.8)
–/80 (–/ 100)
8.6 (5–11)
Pepler [59]
48/27 (64.0/ 36.0)
15.1 (12–17)
–
Type CBT
Diagnoses
Age (in years)
# men/ women (%)
Intervention
Study population
Ogden [58]
References
Table 1 continued
Eur Child Adolesc Psychiatry
123
Eur Child Adolesc Psychiatry
Fig. 1 Forest plot, comparison of effect sizes per study
4 = good, 3 = fair, 2 = poor, 1 = very poor), 13 studies (52 %) were rated as ‘fair’ or higher. Effectiveness of CBT The effect sizes of the conducted studies ranged between d = -0.19 and d = 2.35 with an overall mean weighted effect size of d = 0.50, indicating a medium treatment effect for CBT to reduce aggression in adolescents. Heterogeneity was calculated (see Fig. 1) in RevMan 5.1, using the I-squared and Chi-squared test. Heterogeneity was found in the meta-analysis (I2 = 76 %, v2 = 97.99, p \ 0.0001), which shows inconsistency of study results. Therefore, the random effect model was used to correct for the heterogeneity and the different outcome measurements used in the meta-analysis. Moderators of treatment effect To explain the heterogenic outcome of the meta-analysis, we investigated whether effect sizes were moderated by specific participant and treatment characteristics. A metaregression analysis was conducted to compare the effect
123
sizes of the different studies (see Table 2). Focusing on the participant characteristics, no effect of gender and age was found regarding the treatment effectiveness of CBT. Furthermore, only 10 studies reported clinical diagnosis of the participants. The moderator analysis of these reported diagnoses did not reveal significant effects. Unfortunately, only two studies in our meta-analysis made a distinction between proactive and reactive aggression regarding the treatment of aggression (studies of [36, 37]). Because of this lack of research, proactive or reactive aggression could not be examined as moderators of treatment effect in the meta-analysis. The analysis of intervention characteristics showed that treatment duration did not significantly predict treatment effect. Also, no significant moderator effects of treatment in clinical setting, school setting or home interventions were found. Furthermore, no significant difference between the different intervention types was found (group, individual or family intervention). Of the seventeen studies that reported a first follow-up period after post-treatment, six studies reported second additional follow-up outcomes after treatment. Follow-up-treatment effect sizes ranged between d = -0.04 and d = 3.19 with an overall mean
Eur Child Adolesc Psychiatry Table 2 Values of the moderator analysis Moderators of treatment success
Level of variable (number of studies)
Gender
Boys n = 5
0.49
Girls n = 3
1.089
Boys and Girls n = 17
0.39
ES per level
Age Diagnosis n = 10
0.42 0.54
Setting
No Diagnosis n = 15 Clinic n = 19
0.61
School n = 4
0.098
Home n = 1 Unknown n = 1
F(3,24) = 1.223, p = 0.326
0.49
Group n = 15
0.35
Family n = 9
0.78
Duration of treatment
Mean = 72.94 h
Unknown n = 1
F (1,24) = 0.206, p = 0.655
-0.10
Intervention type
Jadad quality
F(2,24) = 1.916, p = 0.171
F(1,24) = 0.977, p = 0.550
Diagnosis
Rater
p value
F(2,24) = 2.058, p = 0.152
-0.02 F(12,15) = 0.354, p = 0.916
Median = 24 h \24 h n = 8
0.71
C24 h n = 8
0.58
No information n=9
0.22
Open n = 15
0.38
Blind n = 10
0.67
Very poor n = 2
0.70
Poor n = 10
0.38
Fair n = 5
0.29
Good n = 2
0.18
Excellent n = 6
0.89
F(2,24) = 1.646, p = 0.216
F(1,24) = 1.509, p = 0.232 F(4,24) = 1.124, p = 0.373
weighted effect size of d = 0.92, indicating large treatment effects on longer terms. However, follow-up effect sizes were not statistically significant different compared to posttreatment effect sizes (t = 0.42, p = 0.698). Looking at the study differences between open and blinded raters and the quality of the studies according to the Jadad scale (low versus high), no significant differences were found between effect sizes.
Discussion The aim of this study was to examine response predictors of CBT interventions in children and adolescents with aggression problems. Besides the common predictors such as gender, age, diagnosis, setting, treatment delivery and quality of the study, proactive and reactive aggressions were investigated as treatment moderators. A meta-analysis was
conducted to define these moderators. Overall, results of the meta-analysis study showed a medium treatment effect of CBT to reduce aggression in children and adolescents (d = 0.50), which is in line with earlier meta-analysis [14, 15]. The results were heterogeneous, therefore a random effect model was used and moderators of effect were investigated. Looking at participant characteristics as moderators of treatment effect in the meta-analysis, age, gender, diagnosis, type of raters and type of aggression (proactive and reactive aggression) were examined as possible predictors of response to the treatment. None of these moderators were found to predict treatment effects in the meta-analysis. Gender and age were not found as moderators of effect in a previous meta-analysis either [14, 38]. Unfortunately, proactive and reactive aggressions were only mentioned in two of the 25 included studies and therefore could not be examined as a possible predictor of response to treatment. However, the study of Down et al. [36] showed that treatment focused on reactive aggression (CBT) but also on proactive aggression (Personal Developmental Anger Management) was positive related to higher levels of anger control and self esteem after following this treatment. Age was found as a moderator of treatment effect in the reactive treatment group, where young adolescents did show lower levels of anger control and self esteem after following the treatment compared to the older adolescents. Furthermore, Van Manen et al. [37] showed that after following the Social Cognitive Intervention Program and the Social Skills Program, proactive and reactive aggression decreased in 9- to 13-year old boys after 1-year follow-up. However, participants with reactive aggression reacted significantly better after following the Social Cognitive Intervention Program (based on social information processing) than participants with proactive aggression. Since several possible moderators of effect were examined and none differed significantly from each other, the heterogeneity of the meta-analysis might be caused by the lack of differentiation between proactive and reactive aggression in the majority of the studies included in this meta-analyses. Overall, the meta-analysis indicates that the external and in particular predictive validity of proactive and reactive aggression regarding the response to CBT needs to be further investigated. The diagnoses of the participants were only reported in 10 of the 25 studies. No significant effect was found in these 10 studies that reported the diagnosis of the participants (overall ODD/ CD), however this could be due to the fact of lacking data. In future research, studies should report their samples in more detail regarding behavioral symptoms and clinical diagnosis of the participants as this could be a possible moderator variable. For example, it has been shown that reactive aggression is associated with anxiety and attention
123
Eur Child Adolesc Psychiatry
problems and proactive aggression is associated with delinquent behavior and CD [18]. More insight in these possible moderator variables, could lead to more effective and personalized interventions. Finally, treatment characteristics were examined as possible moderators of treatment response. Results indicated that treatment duration, treatment setting (school versus clinical) and intervention type (e.g., individual, group or family) did not predict treatment outcome of CBT. The finding that treatment duration does not seem to influence outcome is quite remarkable, but is in line with results of previous meta-analyses [14, 39]. This may suggest that more cost-effective development of future interventions is feasible. Moreover, there is even some evidence suggesting that recidivism is lower after treatment of shorter duration versus longer duration [40, 41]. This could possibly be explained by that short interventions create a larger momentum and/or are more intensive. The finding that treatment setting also does not seem to influence outcome is hopeful, since it means that less-invasive (and often cost-effective) school-based interventions may be preferred compared to clinical interventions. However, there are studies suggesting that clinical interventions are more effective than school-based interventions in clinical severe cases, suggesting it is premature to prefer schoolbased interventions over clinical interventions in all cases [39]. Lastly, intervention type did not seem to moderate treatment outcome, although a meta-analysis did indicate that including parents in the treatment of adolescents is important [15]. Since the lack of power could have caused the non-significant results of the moderator analysis, future research should examine these possible moderators of treatment effect in larger meta-analyses to develop more effective and cost-beneficial treatment of aggression. Also, the meta-analysis did not find significant difference between effect sizes of studies with open raters or blinded studies, or between that of studies with high versus low quality. However, this could be due to that different outcome measurements were used in the studies included in our meta-analysis. This could lead to a bias in the results, therefore studies with matching outcome measurements are highly needed in future to have a better indication of the overall outcomes regarding the treatment of aggression. Limitations This study had some limitations. First of all, a possible limitation of the present meta-analysis is that the search strategy primarily relied on published studies. As a result, magnitudes of treatment effects might have been overestimated [42]. On the other hand, it has been argued that including grey literature could also lead to a bias of the results, since grey literature often includes non-peer
123
reviewed articles with lower level of methodologically quality [43]. Furthermore, only a small number of studies were included in the moderator analyses, which could have biased the results of the moderator analyses due to a lack of power.
Conclusion The findings of this meta-analysis indicate that CBT is effective in decreasing aggression in adolescents (d = 0.50). Treatment style, age and gender were not found as moderators of response to treatment. Too few studies examined the moderating role of proactive and reactive aggression as predictors of response to treatment, and thus further research is needed in this area. Also, many studies did insufficiently report on clinical diagnosis and behavioral symptoms of the study sample. Furthermore, it was shown that treatment setting and duration did not seem to influence treatment effect, which shows the need for development of more cost-effective and lessinvasive interventions. Finally, better standardization of design, predictors, and outcome measures across studies is required to establish predictive profiles of responders and non-responders to treatment of reducing aggression. Conflict of interest Jan K. Buitelaar has been a consultant to/ member of advisory board of/and/or speaker for Janssen Cilag BV, Eli Lilly, Bristol-Myer Squibb, Shering Plough, UCB, Shire, Novartis and Servier. He is not an employee of any of these companies, and not a stock shareholder of any of these companies. He has no other financial or material support, including expert testimony, patents, royalties. F.E.Scheepers has travel support and is a speaker for Eli Lilly. K. Smeets, A.Leeijen, M. van der Molen and N. Rommelse state no conflict of interest.
References 1. Armbruster MA, Sukhodolsky DG, Michalsen R (2004) The impact of managed care on children’s outpatient treatment: a comparison study of treatment outcome before and after managed care. Am J Orthopsychiatry 74(1):5–13 2. Rutter M, Bishop D, Pine D, Scott S, Stevenson JS, Taylor EA, Thapar A (2010) Rutter’s child and adolescent psychiatry. Wiley, Oxford 3. Gannon TA, Ward T, Beech AR, Fisher D (2007) Aggressive offenders’ cognition: theory, research and practice. Wiley, Chichester 4. Romeo R, Knapp M, Scott S (2006) Economic cost of severe antisocial behaviour in children—and who pays it. Br J Psychiatry 18(8):547–553 5. Buitelaar JK, Smeets KC, Herpers P, Scheepers F, Glennon J, Rommelse NNJ (2013) Conduct disorders. Eur Child Adolesc Psychiatry 22(1):49–54 6. Frick PJ, Viding E (2009) Antisocial behavior from a developmental psychopathology perspective. Dev Psychopathol 21(04): 1111–1131
Eur Child Adolesc Psychiatry 7. Monahan KC, Steinberg L, Cauffman E (2009) Affiliation with antisocial peers, susceptibility to peer influence, and antisocial behavior during the transition to adulthood. Dev Psychol 45(6):1520–1530 8. Armelius BA, Andreassen TH (2007) Cognitive-behavioral treatment for antisocial behavior in youth in residental treatment. Cochrane Database Syst Rev 4:1–35 9. Connor DF, Carlson GA, Chang KD, Daniolos PT, Ferziger R, Findling RL, Hutchinson JG, Malone RP, Halperin JM, Plattner B, Post RM, Reynolds DL, Rogers KM, Saxena K, Steiner H (2006) Juvenile maladaptive aggression: a review of prevention, treatment, and service configuration and a proposed research agenda. J Clin Psychiatry 67(5):808–820 10. Barzman DH, Findling RL (2008) Pharmacological treatment of pathologic aggression in children. Int Rev Psychiatry 20:151–157 11. Mattaini MA, McGuire MS (2006) Behavioral strategies for constructing nonviolent cultures with youth: a review. Behav Modif 30(2):184–224 12. Blake CS, Hamrin V (2007) Current approaches to the assessment and management of anger and aggression in youth: a review. J Child Adolesc Psychiatr Nurs 20(4):209–221 13. Aman MG, Bukstein OG, Gadow KD, LE Arnold, Molina BS, McNamara NK, Rundberg-Rivera EV, Li X, Kipp H, Schneider J, Butter EM, Baker J, Sprafkin J, Rice RR Jr, Bangalore SS, Farmer CA, Austin AB, Buchan-Page KA, Brown NV, Hurt EA, Grondhuis SN, Findling RL (2014) What does risperidone add to parent training and stimulant for severe aggression in child attention-deficit/hyperactivity disorder? J Am Acad Child Adolesc Psychiatry 53(1):47–60. doi:10.1016/j.jaac.2013.09.022 e41 14. Sukhodolsky DG, Kassinove H, Gorman BS (2004) Cognitivebehavioral therapy for anger in children and adolescents: a metaanalysis. Aggress Viol Behav 9(3):247–269 15. Fossum S, Handega˚rd BH, Martinussen M, Mørch WT (2008) Psychosocial interventions for disruptive and aggressive behaviour in children and adolescents: a meta-analysis. Eur Child Adolesc Psychiatry 17:438–451 16. Sonuga-Barke EJ, Brandeis D, Cortese S, Daley D, Ferrin M, Holtmann M, Stevenson J, Danckaerts M, van der Oord S, Dopfner M, Dittmann RW, Simonoff E, Zuddas A, Banaschewski T, Buitelaar J, Coghill D, Hollis C, Konofal E, Lecendreux M, Wong IC, Sergeant J (2013) Nonpharmacological interventions for ADHD: systematic review and meta-analyses of randomized controlled trials of dietary and psychological treatments. Am J Psychiatry 170(3):275–289 17. Polman H, Orobio de Castro B, Koops W, van Boxtel HW, Merk WM (2007) A meta-analysis of the distinction between reactive and pro active aggression in children and adolescents. J Abnorm Child Psychol 35:522–535 18. Vitaro F, Brendgen M, Tremblay RE (2002) Reactively and proactively aggressive children: antecedent and subsequent characteristics. J Child Psychol Psychiatry 43(4):495–505 19. Stanford MS, Houston RJ, Mathias CW, Villemarette-Pittman NR, Helfritz LE, Conklin SM (2003) Characterizing aggressive behavior. Assessment 10(2):183–190 20. Kockler TR, Stanford MS, Nelson CE, Meloy JR, Sanford K (2006) Characterizing Aggressive Behavior in a Forensic Population. Am J Orthopsychiatry 76(1):80–85 21. Frick PJ, Ellis M (1999) Callous-unemotional traits and subtypes of conduct disorder. Clin Child Fam Psychol Rev 2(3):149–168 22. Blair RJ (2001) Neurocognitive models of aggression, the antisocial personality disorders, and psychopathy. J Neurol Neurosurg Psychiatry 71(6):727–731 23. Blair RJ, Peschardt KS, Budhani S, Mitchell DG, Pine DS (2006) The development of psychopathy. J Child Psychol Psychiatry 47(3–4):262–276
24. Vitaro F, Brendgen M, Barker ED (2006) Subtypes of aggressive behaviors: a developmental perspective. Int J Behav Dev 30(1):12–19 25. Barker ED, Vitaro F, Lacourse E, Fontaine NMG, Carbonneau R, Tremblay RE (2010) Testing the developmental distinctiveness of male proactive and reactive aggression with a nested longitudinal experimental intervention. Aggress Behav 36:127–140 26. Haas SM, Waschbusch DA, Pelham WE Jr, King S, Andrade BF, Carrey NJ (2011) Treatment response in CP/ADHD children with callous/unemotional traits. J Abnorm Child Psychol 39:541–552 27. Masi G, Manfredi A, Milone A, Muatori P, Polidori L, Ruglioni L, Muratori F (2011) Predictors of nonresponse to psychosocial treatment in children and adolescents with disruptive behaviour disorders. J Child Adolesc Psychopharmacol 21(1):51–55 28. Card NA, Little TD (2006) Proactive and reactive aggression in childhood and adolescence: a meta-analysis of differential relations with psychosocial adjustment. Int J Behav Dev 30(5): 466–480. doi:10.1177/0165025406071904 29. Poulin F, Boivin M (2000) The role of proactive and reactive aggression in the formation and development of boys’ friendships. Dev Psychol 36(2):233–240 30. Bushman BJ, Anderson CA (2001) Is it time to pull the plug on the hostile versus instrumental aggression dichotomy? Psychol Rev 108(1):273–279 31. Morris SB (2008) Estimating effect sizes from pretest-posttestcontrol group designs. Organ Res Methods 11(2):364–386 32. Lipsey MW, Landenberger NA, Wilson SJ (2007) Effects of cognitive-behavioral programs for criminal offenders. Campbell Syst Rev 6. doi:10.4073/csr.2007.6 33. Jadad AR, Moore RA, Carroll D, Jenkinson C, Reynolds DJ, Gavaghan DJ, McQuay HJ (1996) Assessing the quality of reports of randomized clinical trials: is blinding necessary? Control Clin Trials 17(1):1–12 34. Cohen J (1988) Statistical power analysis for the behavioral sciences. Lawrence Erlbaum Associations, Hillsdale 35. Nordic Cochrane Centre TCC (2011) Review Manager (RevMan). (Version 5.1) 36. Down R, Willner P, Watts L, Griffiths J (2011) Anger Management groups for adolescents: a mixed-methods study of efficacy and treatment preferences. Clin Child Psychol Psychiatry 16(1):33–52 37. Van Manen TG, Prins PJ, Emmelkamp PM (2004) Reducing aggressive behavior in boys with a social cognitive group treatment: results of a randomized, controlled trial. J Am Acad Child Adolesc Psychiatry 43(12):1478–1487 38. Landenberger NA, Lipsey MW (2005) The positive effects of cognitive–behavioral programs for offenders: a meta-analysis of factors associated with effective treatment. J Exp Criminol 1:451–476 39. McCart MR, Priester PE, Davies WH, Azen R (2006) Differential effectiveness of behavioral parent-training and cognitive-behavioral therapy for antisocial youth: a meta-analysis. J Abnorm Child Psychol 34(4):527–543 40. Dekovic M, Slagt MI, Asscher JJ, Boendermaker L, Eichelsheim VI, Prinzie P (2011) Effects of early prevention programs on adult criminal offending: a meta-analysis. Clin Psychol Rev 31(4):532–544. doi:10.1016/j.cpr.2010.12.003 41. van der Put CE, Asscher JJ, Stams GJ, van der Laan PH, Breuk R, Jongman E, Doreleijers T (2013) Recidivism after treatment in a forensic youth-psychiatric setting: the effect of treatment characteristics. Int J Offender Ther Comp Criminol 57(9):1120–1139. doi:10.1177/0306624x12452389 42. Dickersin K (1990) The existence of publication bias and risk factors for its occurrence. J Am Med Assoc 263(10):1385–1389
123
Eur Child Adolesc Psychiatry 43. Hopewell S, McDonald S, Clarke MJ, Egger M (2007) Grey literature in meta-analyses of randomized trials of health care interventions. Cochrane Database Syst Rev 2:1–16 55.
Studies included in the meta-analysis 44. August GJ, Lee SS, Bloomquist ML, Realmuto GM, ektner JM (2003) Dissemination of an evidence-based prevention innovation for aggressive children living in culturally diverse, urban neighborhoods: the Early Risers effectiveness study. Prev Sci 4(4):271–286 45. August GJ, Realmuto GM, Hektner JM, Bloomquist ML (2001) An integrated components preventive intervention for aggressive elementary school children: the early risers program. J Consult Clin Psychol 69(4):614–626 46. Brotman LM, Klein RG, Kamboukos D, Brown EJ, Coard SI, Sosinsky LS (2003) Preventive intervention for urban, lowincome preschoolers at familial risk for conduct problems: a randomized pilot study. J Clin Child Adolesc Psychol 32(2):246–257 47. Butler S, Baruch G, Hickey N (2011) A randomized controlled trial of multisystemic therapy and a statutory therapeutic intervention for young offenders. J Am Acad Child Adolesc Psychiatry 50(1220–1235):e1222 48. Carpenter EM (2002) A curriculum-based approach for socialcognitive skills training: an intervention targeting aggression in Head Start preschoolers. University of Maine 49. Feinfield KA, Baker BL (2004) Empirical support for a treatment program for families of young children with externalizing problems. J Clin Child Adolesc Psychol 33(1):182–195 50. Larsson B, Fossum S, Clifford G, Drugli MB, Handega˚rd BH, Mørch WT (2009) Treatment of oppositional defiant and conduct problems in young Norwegian children : results of a randomized controlled trial. Eur Child Adolesc Psychiatry 18(1):42–52 51. Lipman EL, Boyle MH, Cunningham C, Kenny M, Sniderman C, Duku E, Mills B, Evans P, Waymouth M (2006) Testing effectiveness of a community-based aggression management program for children 7–11 years old and their families. J Am Acad Child Adolesc Psychiatry 45(9):1085–1093 52. Mitchell P, Smedley K, Kenning C, McKee A, Woods D, Rennie CE, Bell RV, Aryamanesh M, Dolan M (2011) Cognitive behaviour therapy for adolescent offenders with mental health problems in custody. J Adolesc Health 34(3):433–443 53. Nickel M, Luley J, Krawczyk J, Nickel C, Widermann C, Lahmann C, Muehlbacher M, Forthuber P, Kettler C, Leiberich P, Tritt K, Mitterlehner F, Kaplan P, Pedrosa Gil F, Rother W, Loew T (2006) Bullying girls—changes after brief strategic family therapy: a randomized, prospective, controlled trial with 1-year follow-up. Psychother Psychosom 75(1):47–55 54. Nickel MK, Krawczyk J, Nickel C, Forthuber P, Kettler C, Leiberich P, Muehlbacher M, Tritt K, Mitterlehner FO, Lahmann C, Rother WK, Loew TH (2005) Anger, interpersonal
123
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
relationships, and health-related quality of life in bullying boys who are treated with outpatient family therapy: a randomized, prospective, controlled trial with 1 year of follow-up. Pediatrics 116(2):247–254 Nickel MK, Muehlbacher M, Kaplan P, Krawczyk J, Buschmann W, Kettler C, Rother N, Egger C, Rother WK, Loew TK, Nickel C (2006) Influence of family therapy on bullying behaviour, cortisol secretion, anger, and quality of life in bullying male adolescents: a randomized, prospective, controlled study. Can J Psychiatry 51(6):355–362 Nickel MK, Nickel C, Leiberick R, Tritt K, Mitterlehner FO, Lahmann C, Rother WK, Loew TH (2005) Aggressive female youth benefit from outpatient family therapy: a randomized, prospective, controlled trial. Pediatri Int 47(2):167–171 O’Neal CR, Brotman LM, Huang KY, Gouley KK, Kamboukos D, Calzada EJ, Pine DS (2010) Understanding relations among early family environment, cortisol response, and child aggression via a prevention experiment. Child Dev 81(1):290–305 Ogden T, Hagen KA (2006) Multisystemic treatment of serious behaviour problems in youth: sustainability of effectiveness 2 years after intake. Child Adolesc Mental Health 11(3):142–149 Pepler D, Walsh M, Yuile A, Levene K, Jiang D, Vaughan A, Webber J (2010) Bridging the gender gap: interventions with aggressive girls and their parents. Prev Sci 11(3):229–238 Rowland MD, Halliday-Boykins CA, Henggeler SW, Cunningham PB, Lee TG, Kruesi MJP, Shapiro SB (2005) A randomized trial of multisystemic therapy with Hawaii’s felix class youths. J Emot Behav Disord 13(1):13–23 Santisteban DA, Coatsworth JD, Perez-Vidal A, Kurtines WM, Schwartz SJ, LaPerriere A, Szapocznik J (2003) Efficacy of brief strategic family therapy in modifying Hispanic adolescent behavior problems and substance use. J Fam Psychol 17(1): 121–133 Schechtman Z (2000) An innovative intervention for treatment of children and adolescent aggression: an outcome study. Psychol Schools 37(2):157–167 Sukhodolsky DG, Golub A, Stone EC, Orban L (2005) Dismantling anger control training for children: a randomized pilot study of social problem-solving versus social skills training components. Behav Ther 36:15–23 Sundell K, Hansson K, Lo¨fholm CA, Olsson T, Gustle LH, Kadesjo¨ C (2008) The transportability of multisystemic therapy to Sweden: short-term results from a randomized trial of conductdisordered youths. J Fam Psychol 22(4):550–560 Webster-Stratton C, Reid J, Hammond M (2001) Social skills and problem-solving training for children with early-onset conduct problems: who benefits? J Child Psychol Psychiatry 42(7): 943–952 van de Wiel NM, Matthys W, Cohen-Kettenis PT, Maassen GH, Lochman JE, van Engeland H (2007) The effectiveness of an experimental treatment when compared to care as usual depends on the type of care as usual. Behav Modif 31(3):298–312
