Incredible Years – Parent Training (IY-PT)

Four RCTs (1 high, 2 moderate, and 1 low risk of bias)^{93,102,129,138} and one prospective cohort study (moderate risk of bias)⁸⁷ evaluated a version of the IY-PT (Table 7). Of these, two RCTs^129,138 and the prospective cohort study⁸⁷ evaluated the standard version of the IY-PT, one RCT⁹³ evaluated a brief version of the IY-PT, and one RCT (reported in 2 publications)^102,159 evaluated a nurse-led or therapist-led version of the IY-PT.

Table 7

Summary of behavior outcomes from studies of a parent-only component (IY-PT) in preschool-age children.

All three of the studies evaluating the IY-PT standard version measured child disruptive behaviors using the ECBI Problem scale.^{87,129,138,163} Two of these studies also used the ECBI Intensity scale and Dyadic Parent-Child Interaction Coding System-Revised (DPICS-R) as additional measures of child disruptive behaviors.^87,138 One study used the Strengths and Difficulties Questionnaire (SDQ)¹³⁸ and one used the CBCL.^102,159 Comparison groups included usual care (n = 1),⁸⁷ waitlist control group (n = 3),^93,129,138 and a group led by a different provider or receiving no interventionist-led training (1 study reported in 2 publications).^102,159 Timing of the final followup ranged from 3 months to 2 years post-intervention across studies. Table 7 summarizes key outcomes. Briefly, in three of the five studies, the groups receiving IY intervention had significantly improved behavioral outcomes compared with control arms. In one study comparing differing administration of the IY intervention and in comparing IY to usual care, ECBI outcomes did not differ significantly among groups.

The moderate risk of bias prospective cohort study⁸⁷ compared outcomes in 4-year old children [mean age: 4.2 (3.11)] scoring at or above the 80^th percentile on the aggressive behavior scale of the CBCL to outcomes in children receiving usual care. Parents of children [n = 72, mean age: 50.3 (3.11) months, mean CBCL aggressive behavior raw score: 21.99 (4.37)] in the intervention group received 18 2-hour IY sessions (BASIC and ADVANCE) focusing on identifying strategies for dealing with child behaviors. Children in the control group [n = 72, mean age: 51.3 (2.53) months, CBCL aggressive behavior mean raw score: 22.49 (4.69)] received usual care. Groups differed significantly at baseline on age (control group 2 months older than intervention group, p=0.02), observed use of critical statements by parents (intervention group parents more critical than control, p=0.05), and observed conduct problems (more conduct problem in intervention group vs. control, p=0.004). Children did not differ at baseline on parent-rated measures. At final followup (2 years post-intervention), groups did not differ significantly on the ECBI. In observer coding (DPICS-R) of interactions, parents in the intervention group used significantly fewer critical statements than in the control group, and conduct problems decreased significantly in the intervention group compared with the control arm.

One RCT (reported in 2 publications)^129,158 compared an IY intervention delivered by center staff in social service centers for economically disadvantaged children in the United Kingdom with a waitlist control group. Children included in the study were seen at the centers and had ECBI Intensity scores of 127 or greater or problem scores of 11 or greater, and most had low socioeconomic status. Eighty-six of the 104 children randomized to the IY group [mean age: 46.4 (6.6) months] and 45 of the 47 randomized to the waitlist control group [mean age: 46.2 (4.2) months] completed the followup assessments at 6 months post-treatment. In intention-to-treat analyses, the outcomes on both ECBI scales (intensity scale effect size: 0.89, 95% CI: 0.54 to 1.24; problem scale effect size: 0.63, 95% CI: 0.28 to 0.98) and on the SDQ conduct problems (effect size: 0.33, 95% CI: −0.02 to 0.68) scale were significantly improved (p values <0.05) in the IY group compared with controls. Instances of deviant child behavior coded in observations were lower in the IY group but group but group differences were not significant (effect size: 0.21; 95% CI: −0.13 to 0.55). Scores on hyperactivity scales (Conners, SDQ) were also significantly lower in the IY group compared with control (p values <0.05), while scores on the SDQ overall deviance scale did not differ significantly between groups.

Another trial conducted in Irish community service centers enrolled 149 children between the ages of 32 and 88 months who were referred to health services organizations for problem behaviors and who scored above the clinical cut offs (127 for Intensity and 11 for problem scale) on a parent-rated ECBI.^138,163 The IY intervention was delivered by center staff. Ninety-five of the 103 children randomized to the IY group and 42 of 45 in the waitlist control group completed followup final assessments approximately 3 months after the end of treatment. In intention-to-treat analyses, the IY group improved significantly on both ECBI scales (p values <0.001) compared with the control arm (ECBI Intensity effect size: 0.70; 95% CI: 0.4 to 1.1; ECBI Problem subscale effect size: 0.75; 95% CI: 0.4 to 1.1). Scores on measures of hyperactivity, prosocial behavior, and emotional well-being were also significantly improved in the IY arm compared with control (p values <0.01). Child problem behaviors coded in observations also decreased significantly in the IY arm versus the control arm (effect size: 1.07, 95% CI: 0.6 to 1.6), but observations of positive child behavior did not differ significantly between groups. Investigators conducted observations with a subset of children in both groups. Intention-to-treat analyses of the children originally randomized to the IY-PT group at 12-month post-treatment followup assessment demonstrated that treatment effects were maintained from 6-month followup (e.g., end of treatment) to 12-month followup, although effect sizes were nominally – but not statistically significantly – smaller.¹⁶³

A moderate risk of bias RCT⁹³ compared outcomes in three groups: an intervention group randomly allocated to 10 weeks of IY parent training [n = 89, mean age: 2.8 (0.61) years]; a non-randomly allocated group receiving the 10-week training [n = 123, mean age: 2.90 (0.63) years]; and a randomly allocated waitlist control group [n = 61, mean age: 2.7 (0.55) years].⁹³ All children had Infant-Toddler Social Emotional Assessment Scale scores at or above the 80^th percentile. Groups were similar at baseline; however, families in the non-random IY group included more minorities and were more likely to report lower socioeconomic status. Mean baseline T-scores on the ECBI Problem subscale ranged from 60.1 to 62.8 and from 58.3 to 59.2 on the ECBI Intensity scale. At the 12-month followup, outcomes on the ECBI Problem and Intensity subscales and the Parenting Scale were significantly improved in both the IY arms compared with the control group (p<0.05). Mean decreases in negative parenting, child disruptive behaviors, and negative parent-child interaction coded on the DPICS-R were greater in the IY groups compared with the control group and did not differ significantly between the IY groups.

Another high risk of bias RCT evaluated outcomes following 6 to 12-week IY programs led by primary care nurses (n = 49 children) or by psychologists (n = 37 children) and among a group of children whose parents received the Incredible Years book but no specific interventionist-led training (n = 31 children).^102,159 While the study enrolled 117 children, only 91 completed all assessments (77%). All children were between the ages of 3 and 6.11 years, and all had scores above the 90^th percentile on the CBCL Externalizing scale and DSM-IV diagnoses of Oppositional Defiant Disorder (ODD). The mean baseline CBCL Externalizing score (SD) across groups was 70.7 (5.96) and mean ECBI Intensity score (SD) was 155.44 (27.41). Groups did not differ demographically or in comorbidities (27.4% with concomitant ADHD). At 12-months post-intervention, groups did not differ significantly on any ECBI or CBCL scale, though all groups improved from baseline. Scores on the ECBI were in the normal range for 23.1 percent of children across groups at followup and were in the normal range for 47.9 percent of children on the CBCL Externalizing scale. In equivalence testing, the combined interventionist-led groups and book-only group were equivalent at the 10 percent level (differing by <10% at post-treatment and the 12-month followup) on both scales, as were the nurse-led and psychologist-led groups. In dose-effect analyses, effects on both scales improved with increasing training sessions attended.

Positive Parenting Program (Triple P)

Five RCTs (2 high and 3 moderate risk of bias) evaluated a version of Triple P.^{135,139-141,145} Two studies evaluated a self-directed version,^139,141 two studies evaluated an enhanced version,^135,140 and one study evaluated an online version (Table 8).¹⁴⁵ All RCTs of Triple P that met criteria for inclusion in this review were conducted in Australia, two in rural populations.

Table 8

Summary of behavior outcomes from studies of a parent-only component (Triple P) in preschool-age children.

All five studies measured child disruptive behaviors with the ECBI Intensity and Problem subscales. Four of the five studies also used the Parent Daily Report (PDR),^135,139-141 one study also used the SDQ,¹⁴⁵ and only one study measured one of our protocol-defined functional outcomes.¹⁴⁰ The only study to use direct observation¹⁴⁵ did not use this measure for all participants. Total comparison groups included waitlist control groups (n = 4) and usual care or self-directed treatment (n = 2). The duration of treatment ranged from 8 to 11 weekly sessions. Timing of last followup was 4 months post-intervention in one study,¹⁴¹ 6 months post-intervention in three studies^135,139,145 and 1 and 3 years post-intervention in one.^140,160,162 Table 8 summarizes key outcomes. Overall, ECBI outcomes and mean number of problem behaviors were improved in treatment arms compared with control.

One RCT comparing a 10-week, self-directed iteration of Triple P implemented by parents in rural areas of Australia to a waitlist control group reported improvements in behavioral outcomes in the intervention group [n = 12, mean age: 49.33 (14.05) months] compared with control [n = 11, mean age: 53.18 (11.26) months].¹⁴¹ Children were initially identified based on parent concern and interest in the study. Parents completed a DSM-IV diagnostic interview by phone to evaluate for the presence of ADHD (present in 5 intervention and 7 waitlist children), ODD (present in 8 intervention and 6 waitlist children), or conduct disorder (present in 1 treatment and 2 waitlist children). Intervention group parents received printed books and workbooks to work through each week and participated in weekly calls (mean duration of 20 minutes; range: 5 to 30 minutes) with a trained therapist to encourage problem-solving skills. At the end of the 10-week intervention, ECBI scores as rated by mothers and fathers were significantly improved in the intervention group compared with the control (p=0.0005). Mothers, but not fathers, also rated the number of problem behaviors as significantly improved (p=0.016) in the intervention group compared with control. At post-treatment 33 percent of children in the intervention arm remained above the clinical cut-off (ECBI score=127) for disruptive behavior, compared with 100 percent remaining in the clinical range in the control arm. At followup of seven of 12 intervention groups 4 months after the end of intervention, post-treatment effects were maintained except for an increase in child problem behaviors that remained lower than the mean baseline level. At followup, three of seven children (43%) remained in the clinical range for disruptive behavior on the ECBI.

Another RCT in a rural population compared a similar self-directed version of Triple P plus weekly phone conferences [n = 14, mean age: 47.21 (10.19) months] with self-directed Triple P alone [n = 15, mean age: 47.27 (9.84) months] and with a waitlist control arm [n = 12, mean age: 46.17 (13.29) months].¹³⁵ Children had to have an ECBI Intensity score of ≥127 or problem score of ≥ 11 at baseline and parental concern about disruptive behavior. ECBI Problem and Intensity subscale scores and mean number of problem behaviors as rated by mothers were significantly improved in the treatment groups versus control, with significantly greater improvements in the self-directed plus phone arm compared with either other arm (all p values <0.01). Father-rated measures were not significantly different among groups. At followup 6 months after the end of treatment, effects were maintained for the intervention plus phone group and the intervention alone group, with continuing mother-reported improvements in the level of disruptive behaviors in the latter group. Almost 70 percent (69%, n = 9) of the intervention plus phone group and 57 percent (n = 8) of the intervention alone group showed reliable change on the ECBI Intensity scale at the 6-month followup.

Another RCT of self-directed Triple P [n = 32 at baseline, mean age: 42.91 (9.16) months] compared with a waitlist condition [n = 31 at baseline, mean age: 43.26 (9.10) months] reported similarly improved outcomes after the 10-week intervention in the treatment arm.¹³⁹ Children had to have ECBI Intensity scores of ≥127 or problem score of ≥11 at baseline and parental concern about disruptive behavior. Scores on the ECBI Intensity and problem scales and the mean number of problem behaviors reported by parents were significantly improved in the treatment group (n = 21 at analysis) compared with control (n = 22 at analysis) at post-treatment (all p values <0.01; significance maintained in intention-to-treat analyses). At followup of 13 children in the intervention group 6 months after the end of treatment, improvements in child behavior were maintained, and 23 percent of children (3/13) showed clinically reliable behavioral improvements.

One RCT (reported in multiple publications) evaluated three variations of Triple P in 3-year old children compared with a waitlist control group (n = 77): self-directed alone (n = 75); self-directed plus 10 hours of therapist-led skills training with observation and feedback (n = 77); and self-directed plus 14 hours of skills training that included training in partner support and observation and feedback (n = 76).^140,160,162 Families included in the study had at least one indicator of “family adversity,” which included maternal depression, low socioeconomic status or low occupational prestige, relationship conflict, or single parent family, and all children [mean age: 3.4 (3.66)] scored in the clinical range on the ECBI Intensity (≥127) or problem (≥11) scales. Attrition over the course of the intervention was significant, with 30 percent (66/228) not completing either the post-intervention or 1-year followup assessments. Analyses of attrition indicated that negative affect ratings were higher among parents who did not complete the intervention and that mothers who did not complete the intervention were more likely to rate child behavior negatively. At followup after 15 weeks of intervention, children in the Triple P plus 14-hour training condition had improved outcomes on the ECBI compared with children in the waitlist or self-directed alone conditions. Fathers of children in either of the conditions that included additional training reported fewer behavior problems compared with fathers of children in the waitlist arm. Fewer negative child behaviors were recorded in observations in both the additional training arms compared with the waitlist control (p values <0.05) in the 14-hour training arm compared with the self-directed arm (p<0.05). At the 12-month post-treatment followup, improvements in child behavior were maintained in the arms with additional training, but differences were not significant. Forty children in the 14-hour training arm, 56 in the 10-hour training arm, 32 in the self-directed only arm, and 21 in the waitlist arm had moved from the clinical range for disruptive behavior to the typical range on the ECBI (differences between all treatment groups and waitlist group significant at p<0.01; differences between the 14-hour and self-directed group significant at p<0.05; differences between the 10-hour arm and self-directed and 14-hour and 10-hour arm not significant). In a followup of 139 participants 3 years after the end of intervention,¹⁶⁰ children continued to improve on measures of problem behavior from baseline but differences among groups were not significant, nor were the numbers of children who met diagnostic criteria for disruptive behavior disorders (range 23.4 to 32% with DBD diagnoses across treatment groups). Teacher ratings of behavior problems also did not differ among groups, and all ratings were in the non-clinical range.

A sub-analysis of 87 children with ADHD included in this RCT¹⁶² assigned to the 14-hour training arm (n = 26), the 10-hour arm (n = 29), or the waiting list (n = 32) had similar outcomes to those in the larger group, with significantly improved behaviors rated on the ECBI in children in the treatment arms compared with control at post-intervention. The mean number of problem behaviors was similarly lower in the treatment arms (all p values <0.05), and differences between the two treatment arms were not significant. Effects were maintained at the 1-year followup with no significant differences between treatment arms. At least 60 percent of children in each treatment arm met criteria for reliable change on the ECBI (p=NS) at the 1-year followup.

Finally, one RCT compared an online version of Triple P (n = 60) with internet use as usual (n = 56) among parents of children ages 2 to 9 [mean: 4.7 (1.76)] years with elevated ECBI scores.¹⁴⁵ As in the other studies of Triple P, problem behaviors on the ECBI were significantly reduced in the intervention group compared with control at the 6-month followup (ECBI Problem subscale effect size: 0.60; ECBI Intensity subscale effect size: 0.74). Overall SDQ ratings were not significantly different between groups, nor were observed child disruptive behaviors (effect size: 0.14). At least 60 percent of children in the treatment arms were considered clinically improved on the ECBI Problem (60%, n = 34/57) and Intensity subscales (65%, n = 34/52) at the post-treatment assessment compared with 29 (n = 14/49) and 17 (n = 8/46) percent in the control group (p=0.001).

Other Interventions With Only a Parent Component

Four RCTs conducted in the United States (high risk of bias),⁹⁵ Canada (high risk of bias),¹²⁷ Israel (moderate risk of bias),¹¹⁹ and Australia (moderate risk of bias)¹⁵⁵ examined other interventions including only a parent component (Table 9).^{95,119,127,155} All interventions targeted parent behaviors related to communication and discipline. One study incorporated technology to enhance the Helping the Noncompliant Child intervention,⁹⁵ one adapted parent-training modalities,¹¹⁹ one evaluated Supportive Expressive Therapy-Parent Child model,¹²⁷ and one randomized children to group program for parents called Tuning into Kids (TIK).¹⁵⁵ Comparison groups included IY parent training, minimal intervention, standard, non-enhanced care, and treatment as usual. The final followup occurred at 6 months in one study,¹⁵⁵12 months after the end of intervention in two studies,^119,127 and was not clearly reported in one.⁹⁵

Table 9

Summary of behavior outcomes from studies of a parent-only component (other) in preschool-age children.

One RCT compared the standard, clinic-based Helping the Noncompliant Child program [n = 8, mean age: 5.75 (2.12) years], which emphasizes parental attention and positive parent-child communication and relationships, with a technology-enhanced version [n = 7, mean age: 5.57 (1.27) years] that included the standard clinic-based training plus smartphones to watch video training, complete skill surveys, and to record interactions for feedback.⁹⁵ Both groups were lower income and had ECBI scores in the clinical range (127 on the intensity scale or 11 on the problem scale), with higher intensity scores in the enhanced intervention group compared with standard intervention [148.86 (22.51) vs. 131.5 (2.87), p=NR]. At the end of intervention (timing of followup after the 8-10 intervention sessions per group not clear), scores on the ECBI Intensity and Problem scales favored the enhanced group versus the standard intervention group with between-group effect sizes of 0.99 (95% CI: −0.13 to 2.05) for intensity and 0.54 (95% CI: −0.51 to 1.56) for the problem scale. Pre-post effect sizes for each group were more than 1.0, and post-scores on both ECBI scales were in the normative range for children in the technology-enhanced arm. Post-scores on the ECBI Intensity scale, but not the problem scale, were in the normative range for the standard treatment group.

Another RCT evaluated an intervention program (Hitkashrut) combining elements of parent training models including parental self-regulation, involvement of fathers, parent-child communication skills, and behavior management compared with undefined minimal intervention.¹¹⁹ Children were eligible for the study if they scored in the clinical or sub-clinical range on a teacher-rated SDQ. Behavior outcomes on the ECBI for children in the treatment arm (n = 140, mean age: 48.51 (7.35) months] were significantly improved (p<0.001, effect size: 0.76) at 1 month post-intervention compared with the control arm [n = 69, mean age: 48.62 (6.59) months]. At followup of 60 percent of participants (96 in intervention group, 29 in control) 1-year post-intervention, conduct problems were significantly reduced from baseline in the Hitkashrut arm (p<0.001) but not in the control group. The odds of reliable improvement in conduct problems were higher in children in the treatment arm than for those in the control arm (OR=5.09; 95% CI: 2.14 to 12.11) as were the odds of greater improvement in conduct problems from baseline (OR=3.24; 95% CI: 1.30 to 8.02).

One RCT compared the Supportive Expressive Therapy – Parent Child model (n = 27) to the IY-PT program (n = 27).¹²⁷ Supportive Expressive Therapy entails recognizing and adapting dysfunctional parent responses and expectations for child behavior. Both interventions were conducted among children [mean age: 4.2 (0.96)] referred to an outpatient clinic for externalizing behavior disorders, and groups differed at baseline with significantly greater ECBI and CBCL-rated disruptive behaviors in the IY group compared with control (p=0.013). Outcomes at post-intervention among the 18 treatment group completers and 19 IY completers were improved from baseline in both groups with no significant group differences and mean within-groups effect sizes of 0.66 (0.65) and 1.06 (1.57), respectively (p=NS). Observed child negative behaviors decreased over time in both groups, but group differences were not significant. Seven children in the Supportive Expressive Therapy group and 10 in the IY group no longer met the cut-off for disruptive behaviors in the ECBI or CBCL (exact cut-off used not reported) at post-intervention (p=NS). Improvements in outcomes were maintained at the 1-year followup with no group differences. Eight children in the Supportive Expressive Therapy group and six in the IY group were functioning in the normative range at the 1-year followup (p=NS).

Lastly, one moderate risk of bias RCT compared Tuning into Kids (TIK), described as a 6-week long group program for parents of preschool children, against waitlist clinical treatment as usual control.¹⁵⁵ The study sample included 54 children (78% boys) with a mean age of 59.31 (7.38) months. All children had elevated scores on the parent-reported ECBI Intensity score at baseline, with a mean Intensity Score of 169.34 (2.99) in the intervention group and 165.99 (28.82) in the control group. At end of treatment, scores in both groups decreased to 141.26 (23.79) and 157.46 (31.30) for the intervention and TAU groups, respectively (although the group-by-time interaction was not statistically significant). Similar results were reported for the ECBI Problem score [intervention baseline: 23.14 (5.51), end of treatment: 16.86 (6.66); TAU group baseline: 21.00 (8.26), end of treatment: 20.27 (9.04)]. Teacher-rated behavior intensity and problems via the Sutter-Eyberg Student Behavior Inventory were not measured at the end of treatment. At 6-month followup and with two booster sessions after the initial 6-week intervention, mean parent-rated behavior intensity was 148.61 (32.25) and 148.69 (30.36) and behavior problems was 15.57 (9.44) and 16.25 (9.09) for the intervention and control groups, respectively. At 6-month followup, the mean teacher-rated behavior intensity were 101.12 (35.57) and 137.11 (55.39) and behavior problems were 3.94 (6.50) and 10.12 (9.78), for intervention and treatment as usual groups, respectively. The group-by-time interaction was not reported.

Parent Child Interaction Therapy (PCIT)

Seven RCTs (reported in 10 publications)^{98,99,109,112,114,133,153,156,157,161} evaluated a version of PCIT (Table 10). PCIT focuses on improving parent-child interactions to improve disruptive behaviors and combines child-directed play therapy and parent training in behavior management. Studies were conducted in the United States^{98,99,109,112,114,156,157} and Australia.^133,153,161 Five studies (3 high and 2 moderate risk of bias) evaluated a standard version of PCIT.^{98,109,112,114,153,156} One low risk of bias study evaluated a culturally modified version,^99,157 and one study (moderate risk of bias) evaluated an abbreviated version.^133,161 Comparison groups included treatment as usual, waitlist control, and alternate versions of PCIT. All of the RCTs measured child disruptive behaviors with the ECBI Intensity subscale and five used a version of the DPICS observation coding system. Five of seven studies also used the ECBI Problem subscale,^{98,99,109,112,114} and three of the seven studies used the CBCL externalizing scale.^98,99,157 Last followup after the end of treatment ranged from 4 months to a mean of 15.90 months.

Table 10

Summary of behavior outcomes from studies of multicomponent intervention (PCIT) in preschool-age children.

Five RCTs (3 high and 2 moderate risk of bias) compared standard PCIT intervention with a waitlist control group.^{98,109,112,153} The first RCT assessed outcomes in children born premature and exhibiting externalizing behavior problems.^98,156 Most children were referred to the study by clinical personnel (6% were self-referrals by mothers), and children had to score above the clinically significant range on the CBCL (T-score ≥60) to participate. Fourteen children were randomized to immediate PCIT [mean age: 39.7 (14.2) months] and 14 to the waitlist [mean age: 36.5 (13.0) months]. At the end of treatment assessment (4 months), CBCL and ECBI scores were significantly improved for the PCIT group compared with control (p<0.01). Changes were considered clinically significant (meeting magnitude for reliable change and CBCL T-score <60) in all children in the PCIT group (n = 11 at end of treatment), and in four of 14 children in the waitlist arm. At followup of 10 children in the PCT group 8 months after treatment, eight children maintained clinically significant changes and nine demonstrated continued improvement in behaviors from baseline to the final followup.

Another high risk of bias RCT assessed standard PCIT therapy (n = 37 families) compared with a waitlist control group (n = 27 families) in children with ODD [mean age: 59.2 (12.4) months].¹⁰⁹ Sixty-six percent of the children in the study had concomitant ADHD and 22 percent had conduct disorder. Groups differed at baseline on parental IQ, with higher maternal and paternal IQs among parents in the PCIT group compared with control (p<0.05). At the end of intervention (mean 13 sessions over 4 months), scores on the ECBI Intensity and ECBI Problem scales were significantly improved among the 22 families remaining in the PCIT group compared with the 20 remaining in the control group (p<0.05). At 4 months after the end of intervention treatment gains in the PCIT group were maintained, but the study did not assess within- or between-group differences. An earlier paper¹¹² reporting preliminary data on 50 of the 64 families described in the aforementioned paper¹⁰⁹ also reported greater improvement in the PCIT group as compared to the waitlist control group on both ECBI scales.

A high risk of bias RCT conducted in Australia randomized families to PCIT or waitlist control.¹⁵³ The study sample consisted of 34 children with a mean age of 46.52 (6.83) months in the PCIT group and 46.76 (7.50) months in the waitlist control group. Children met diagnostic criteria for ODD and had a disruptive behavior for at least 6 months. Parents were self-referred to participate. Authors reported pre- and post-intervention symptoms measured by ECBI Intensity scale for the PCIT and waitlist control groups and 6-month post-treatment effects for the PCIT group. The mean ECBI Intensity scores from baseline to end of treatment decreased in both the PCIT [baseline: 166.58 (18.93), end of treatment: 125.24 (21.67)] and waitlist control groups [baseline: 173.82 (22.72), end of treatment = 148.35 (19.05)]. Importantly, the mean scores for the PCIT group were in the normal range at end of treatment but the waitlist control group was not (and the difference was statistically significant).

One moderate risk of bias RCT compared children assigned to receive PCIT or to a waitlist control group. The study sample consisted of 30 children with a mean age of 4.53 (0.90) years. Mother ratings from baseline to end of treatment showed greater decrease on the ECBI Intensity scale for the PCIT group [baseline mean: 173(29.5), end of treatment mean: 133(37.7)] as compared to the waitlist control group [baseline mean: 176 (30.2), end of treatment mean: 170 (36.0)] and the ECBI Problem scale [PCIT baseline mean: 23 (5.8), end of treatment mean: 11 (10.7); WLC baseline mean: 25 (5.4), end of treatment mean: 24 (7.5)]. Similar results are reported for the ECBI problem scale. Two RCTs reported on adapted versions of PCIT—one culturally adapted for Mexican-American children (low risk of bias)^99,157 and one adapted to include self-directed methods to abbreviate treatment (moderate risk of bias).^133,161 The first RCT compared three conditions: standard PCIT [n = 19, mean age: 48.9 (92) months]; PCIT culturally adapted for Mexican-Americans by using cultural references and representations [n = 21; mean age (SD): 54.3(11.6 months]; and treatment as usual, which included cognitive-behavioral therapy (CBT) and family therapy [n = 18, mean age: 55.1 (15.3) months].^99,157 Children included in the study were being treated for behavior problems and had a score above the ECBI clinical cut point (more than 127 on Intensity or more than 11 on Problems scale). Overall, 57 percent of families (n = 33) completed the full course of treatment, and 93 percent (n = 54) completed the post-treatment assessments. At the immediate post-treatment assessment, problem behaviors measured on the ECBI and CBCL were significantly reduced in both the PCIT groups compared with treatment-as-usual, but differences between the PCIT arms were not significant. Children improved on the ECBI from baseline in all three groups (ECBI Intensity scale post-treatment effect sizes: 3.38 in adapted PCIT, 2.14 in PCIT, and 1.78 in control; ECBI Problem scale effect size: 2.84 for adapted PCIT, 1.96 for PCIT, and 1.78 for control). Effect sizes at post-treatment on the CBCL were similarly greater than 1 in the PCIT groups and 0.83 in the control group. Outcomes on observational measures of parent and child-led play and compliance were similarly significantly improved in the PCIT groups compared with control, but no different between PCIT arms. Immediately post-treatment, children in the PCIT groups were below the normative mean for behavior problems on the ECBI Intensity scale and CBCL Externalizing scale, and control children were below the clinical cut-offs. At long-term followup of an unstated number of participants at a mean of 15.90 (4.25) and range 6.58 to 24.47 months after the end of treatment, improvements in problem behaviors were largely maintained, with effect sizes on ECBI and CBCL scales ranging from 0.88 to 3.27 across groups and the largest effect sizes in the adapted PCIT group. Differences between groups were not significant in corrected comparisons at long-term followup, although in uncorrected comparisons, behavior outcomes in the adapted PCIT arm were significantly improved compared with the control arm and did not differ from the standard PCIT group.

The second RCT (moderate risk of bias) compared standard PCIT (n = 17 at analysis); an abbreviated version incorporating videotaped trainings (n = 20 at analysis); and a waitlist control group (n = 17 at analysis) in 54 children with a mean age of 46.75 (6.63) months.^133,161 Children in the study had to score in the clinical range (≥132) on the ECBI Intensity scale, meet DSM-IV criteria for ODD, and have been referred for treatment for disruptive behaviors of 6 months or longer duration. The standard PCIT intervention was delivered over 15.5 hours while the abbreviated version was delivered in 9.5 hours. Immediately post-treatment, mother-rated ECBI scores were significantly reduced in both PCIT arms (with no significant differences between the PCIT groups) compared with the waitlist control group (p<0.01). Mothers in the standard PCIT arm also rated problem behaviors in the home as significantly reduced post-treatment compared with the waitlist (p<0.05), but such differences were not seen in the abbreviated PCIT arm. Differences were only significant for father-reported ECBI scores between the abbreviated PCIT arm and the waitlist (p<0.05). Group differences on the CBCL were not significant, nor were reports of observations of child problem behaviors. In corrected comparisons, however, no comparisons of parent-rated measures were significant. Treatment gains were maintained at 6-month and 12-month post-treatment followup, with no significant differences between the PCIT groups. At a final, 2-year followup of 10 children in the standard PCIT group and an unstated number in the abbreviated group, group differences continued to be nonsignificant, and mean ECBI scores for all children were in the non-clinical range (mother-rated ECBI effect size: −0.24; father-rated ECBI effects size: −0.21). Fifty-six percent of children in the standard PCIT arm and 68 percent in the abbreviated group continued to meet criteria for ODD, and 67 to 70 percent in each arm met reliable change criteria for reduction in mother-rated oppositional behavior (group differences not significant).

Other Multicomponent Interventions

Two RCTs (both high risk of bias)^107,154 evaluated a multicomponent intervention for the treatment of disruptive behavior in preschool-age children (Table 11).

Table 11

Summary of behavior outcomes in studies of other multicomponent interventions in preschool-age children.

One multicomponent intervention RCT¹⁰⁷ evaluated child and maternal outcomes at five time points across 16 months for multiple child and parental interventions. Study participants were 36 children [mean age: 5.67 (1.88) years] with a DSM diagnosis of ODD or conduct disorder whose mothers had sought refuge in a battered women's shelter. Investigators randomly assigned mother-child subject pairs to the intervention group, which received weekly sessions following discharge from a women's shelter and continuing for 8 months. Children and mothers in the intervention group received individualized counseling; mothers also received training in child management skills. Comparison mother-child subject pairs were encouraged through monthly meetings or phone calls to use existing community or shelter services. Groups were similar at baseline for demographic variables and screening measures. Mean CBCL Externalizing scale score at baseline was 66.28 (10.00) in the treatment group compared with 65.56 (9.13) in the comparison group. The treatment group demonstrated a greater rate of decrease in parent reported disruptive behaviors at the third assessment. CBCL score differences between groups were not significant, but the rate of improvement of problems was greater among children in the intervention group compared with the control group. At the fifth assessment (ending the 16-months) the mean CBCL Externalizing scale score was 49.79 (9.17) in the treatment group and 58.59 (13.62) in the comparison group (p=NR). Children in the intervention group (n = NR) moved into the normative range [i.e., less than one standard deviation above the mean CBCL Externalizing T-score for normative group of 50 (10)] on the CBCL Externalizing scale. Control children remained in the clinical range. By the final assessment, 3 of 18 children in the intervention group and 8 of 18 in the control group (p<0.05) had externalizing problems at clinical levels (vs. 13/18 in each group at baseline).

In a second RCT,¹⁵⁴ parents were randomized to behavioral family therapy intervention (BFI) or an enhanced group receiving cognitive therapy in addition to the family behavior therapy intervention (CBFI). Both interventions are described as involving both parents and children, although parents are also described as the primary focus of each intervention. Both interventions include teaching a range of positive parenting techniques and strategies for managing misbehavior. The CBFI intervention also includes cognitive therapy components to treat maternal depression. The study sample included 47 families (mean age of children at intake: 4.39 years). All children were diagnosed with either conduct disorder or ODD either alone or in combination with ADHD. There were no significant group-by-time interactions for any parent-reported (CBCL, PDR) or observational measure (Family Observation Schedule), although significant main effects for time were reported for parent reports of child disruptive behavior via CBCL and PDR measures.

Incredible Years – Parent Training (IY-PT)

Three RCTs (1 high and 2 moderate risk of bias) evaluated a version of the IY-PT (Table 15). Two studies examined IY-PT compared to a waitlist control;^118,130 one study examined IY-PT + ADVANCE compared to the standard IY-PT program.¹¹³ All three studies measured child disruptive behaviors using the ECBI Problem subscale, and two studies also used the ECBI Intensity subscale^118,130 and direct observation of child behaviors (but different behavioral observation coding strategies),^113,130 and one study used each of the Sutter-Eyberg Child Behavior Inventory-Revised, Intensity subscale,¹¹⁸ the SDQ,¹¹⁸ and CBCL subscales.¹¹³

Table 15

Summary of behavior outcomes for studies of parent-only intervention (IY-PT) in school-age children.

An RCT conducted in Sweden examined 4 to 8 year old children (85% boys) referred for outpatient child and adolescent psychiatry services meeting diagnostic criteria for ODD. Patients were randomized to IY-PT or waitlist control group.¹¹⁸ From baseline to the end of active treatment the children randomized to the IY-PT group showed significantly more improvement on the parent reported ECBI Intensity subscale [baseline mean: 160.0 (20.3); end of active treatment mean: 128.6 (26.5); change: 20% reduction] and Problem subscale [baseline mean: 20.83 (4.17); end of active treatment mean: 11.13 (7.85); change: 47% reduction] than did children in the waitlist control group [ECBI-I baseline mean: 152.9 (23.6); ECBI-I end of active treatment mean: 147.1 (26.0); change: 4%; ECBI-P baseline mean: 20.41 (6.58); ECBI-P end of active treatment mean: 17.53 (8.01); change: 14% reduction]. Significant differences were not reported on the SECBI-R Intensity subscale or the SDQ.

An RCT conducted in the United Kingdom examined 2 to 9 year old children (74% boys) referred to outpatient services for conduct problems and scoring above the clinical cutoff on the ECBI Problem subscale.¹³⁰ Children receiving IY-PT experienced greater reductions from baseline to post-intervention (14 weeks) on the ECBI Problem subscale [baseline mean: 20.8 (6.5), post-intervention mean: 12.4 (7.8); change: 40% reduction] ECBI Intensity subscale [baseline mean: 152.7 (39.2), post-intervention mean: 130.7 (29.9), change: 14% reduction], and negative behaviors as measured by direct observation [baseline mean: 58.5 (50.6), post-intervention mean: 30.3 (28.6), change: 48% reduction] as compared to children referred to a waitlist control group [ECBI-P baseline mean: 20.3 (7.0), ECBI-P post-intervention mean: 16.3 (8.6), change: 20% reduction; ECBI-I baseline mean: 156.1 (32.9), ECBI-I post-intervention mean: 148.5 (34.7), change: 5% reduction; observed negative behavior baseline mean: 39.9 (37.0), observed negative behavior post-intervention mean: 35.5 (31.5), change: 11% reduction).

One RCT conducted in the United States randomized 3 to 8 year old children (74% boys) scoring above the clinical cutoff on the ECBI, and meeting diagnostic criteria for ODD to receive either IY-PT plus ADVANCE (which includes videotape modeling plus therapist-led discussion focused on family communication, problem solving, and coping skills) or IY-PT.¹¹³ Although main effects for time are reported for both mother-reported child disruptive behaviors as measured by the ECBI Problem Score [IY-PT ADVANCE baseline mean: 17.04 (7.02), IY-PT ADVANCE post-treatment mean: 10.08 (7.95), IY-PT ADVANCE short-term followup mean: 9.23 (7.10); IY-PT baseline mean: 15.55 (7.71), IY-PT post-treatment mean: 9.52 (5.94), IY-PT short-term followup mean: 6.79 (4.82)], CBCL Behavior Problems subscale [IY-PT ADVANCE baseline mean: 66.21 (8.97), IY-PT ADVANCE post-treatment mean: 58.58 (10.12), IY-PT ADVANCE short-term followup mean: 57.48 (11.05); IY-PT baseline mean: 64.09 (8.55), IY-PT post-treatment mean: 57.82 (9.60), IY-PT short-term followup mean: 55.94 (8.69)], and CBCL Social Competence subscale [IY-PT ADVANCE baseline mean: 38.48 (10.28), IY-PT ADVANCE post-treatment mean: 45.40 (14.47), IY-PT ADVANCE short-term followup mean: 45.76 (10.73); IY-PT baseline mean: 38.00 (12.58), IY-PT post-treatment mean: 43.06 (13.54), IY-PT short-term followup mean: 40.42 (10.76)] indicating significant improvement from baseline to post-treatment and short-term followup on each measure, there were no significant effects for group or the group-by-time interaction indicating no differences between groups or between groups over time. Trends for father-reported outcomes were similar to those for mother-reported outcomes for each of these measures.

Parent Management Training – Oregon Model (PMTO)

Two RCTs (1 high and 1 moderate risk of bias) examined PMTO.^117,125 One study used the CBCL¹²⁵ and the other study the ECBI¹¹⁷ as its primary measure of child disruptive behaviors (Table 16).

Table 16

Summary of behavior outcomes for studies of parent-only intervention (PMTO) in school-age children.

One RCT randomized 4 to 12 year old children (80% boys) in Norway to receive either PMTO or treatment as usual.¹²⁵ Referrals were made through the normal process at the participating children's services agencies. Children assigned to PMTO experienced statistically significant greater reductions from baseline to post-treatment (11 to 12 months post-baseline) in parent-reported child disruptive behaviors as measured by the CBCL Externalizing scale [PMTO baseline mean: 66.44 (9.09), post-treatment mean: 59.69 (9.44); treatment as usual baseline mean: 65.61 (10.75), post-treatment mean: 61.22 (9.85)], but not as measured by the Parent Daily Report (PDR). No treatment main effect was reported for teacher-reported child disruptive behaviors as measured by the TRF. No treatment main effect was reported for observed child disruptive behavior.

One RCT compared a brief version of PMTO to treatment as usual with 3 to 12 year old children (69.1% boys) whose parents contacted a primary care agency due to disruptive behaviors.¹¹⁷ Results indicated that the brief version of PMTO was more effective than treatment as usual from pre- to post-treatment on parent-reported ECBI, Intensity [PMTO baseline mean: 124.94 (27.57), PMTO post-treatment mean: 106.06 (27.80); treatment-as-usual baseline mean: 124.76 (28.42); treatment-as-usual post-treatment mean: 114.43 (28.79)], ECBI, Problem [PMTO baseline mean: 15.45 (7.16), PMTO post-treatment mean: 9.79 (7.57); treatment-as-usual baseline mean: 15.02 (7.40); treatment-as-usual post-treatment mean: 11.64 (7.88)], and Merrell externalizing subscale [PMTO baseline mean: 74.17 (19.67), PMTO post-treatment mean: 64.56 (17.95); treatment as usual baseline mean: 73.72 (19.84); treatment-as-usual post-treatment mean: 68.58 (19.20)]. No significant group-by-time interactions for child disruptive behaviors as measured by teacher-reported Merrell externalizing subscale were reported.

Other Interventions With Only a Parent Component

In addition to the IY-PT and PMTO studies discussed above, six other studies examined interventions including only a parent component.^{88,90,91,121,122,147} Three of these six studies measured child disruptive behaviors with the SDQ,^88,91,122 two studies used the CBCL Externalizing subscale,^90,147 one study used each of the KSADS,¹²¹ DBRS-R,¹²¹ Ohio Scales subscales,⁹⁰ and ECBI Intensity and Problem Scales (Table 17).¹⁴⁷

Table 17

Summary of behavior outcomes for studies of parent-only intervention (other) in school-age children.

One RCT randomized 80 children in Nova Scotia, Canada between the ages of 3 and 7 years (78% boys) with ODD to receive the Parenting the Active Child intervention or treatment as usual.¹²¹ The primary outcome was the percentage of children no longer meeting formal criteria for a KSADS-confirmed ODD diagnosis. In comparison to treatment as usual, children with ODD randomized to receive Parenting the Active Child were significantly less likely to meet ODD diagnostic criteria at 120- and 240-days post-treatment, but were not statistically less likely to meet ODD diagnostic criteria at 365-days post-treatment (percentages by group at each time point were not given). DBRS-R scores were not reported.

One RCT compared a practitioner-directed parent management training program (PMT-P) with a self-directed parent management training program (PMT-S) and waitlist control group in a population of 3 to 10 year old children (60% boys) referred to outpatient clinics in Sweden for disruptive behaviors.¹²² Active treatment was 11 weeks long. Six-month followup data are also provided. In comparison to the children in the waitlist control group, children in both PMT groups experienced statistically significantly greater reductions in parent-reported child disruptive behaviors as measured by the PDR [PMT-P baseline mean: 9.4 (3.8), PMT-P post-treatment mean: 6.0 (4.0), PMT-P 6-month followup mean: 5.0 (3.2); PMT-S baseline mean: 9.7 (3.7), PMT-S post-treatment mean: 7.6 (3.7), PMT-S 6-month followup mean: 6.4 (3.9); WLC baseline mean: 10.6 (3.9), WLC post-treatment mean: 10.1 (4.9), WLC 6-month followup data not reported], ECBI-I [PMT-P baseline mean: 137.5 (20.6), PMT-P post-treatment mean: 118.9 (25.6), PMT-P 6-month followup mean: 115.3 (25.1); PMT-S baseline mean: 137.0 (28.1), PMT-S post-treatment mean: 122.3 (30.8), PMT-S 6-month followup mean: 113.7 (29.7); WLC baseline mean: 140.2 (29.8), WLC post-treatment mean: 139.8 (28.9), WLC 6-month followup data not reported], and ECBI-P [PMT-P baseline mean: 15.5 (5.0), PMT-P post-treatment mean: 10.0 (6.9), PMT-P 6-month followup mean: 8.2 (5.9); PMT-S baseline mean: 15.2 (6.9), PMT-S post-treatment mean: 12.0 (7.5), PMT-S 6-month followup mean: 10.2 (7.1); WLC baseline mean: 16.4 (6.4), WLC post-treatment mean: 16.4 (6.5), WLC 6-month followup data not reported]. No differences were reported in child functional outcomes as measured by the parent-reported Social Competence Scale. Direct comparisons of the PMT-P and PMT-S showed significant between-group effects in favor of PMT-P for child disruptive behaviors as measured by the PDR and ECBI-P (but not the other measures) at post-treatment and that this advantage was stable over the 6-month followup period.

One study using a sequential block design to assign parents of 6 to 11 year old children (80% boys) with disruptive behaviors to the Parents Plus Children's Program (PPCP) or to treatment as usual in outpatient mental health services in Ireland.⁸⁸ Duration of active treatment was 8 weeks long. In comparison to children receiving treatment as usual, children assigned to the PPCP program experienced greater reductions in parent-reported child disruptive behaviors as measured by the SDQ total difficulties score [PPCP baseline mean: 21.19 (6.15), PPCP post-treatment mean: 18.12 (6.23); TAU baseline mean: 22.34 (7.33), TAU post-treatment mean: 22.15 (8.30)] and conduct problems score [PPCP baseline mean: 5.07 (2.06), PPCP post-treatment mean: 3.92 (1.61); TAU baseline mean: 5.28 (2.12), TAU post-treatment mean: 5.53 (2.46)].

One study assigned parents of children (83% boys) with ODD (mean age: 9.5 years) referred to a mental health clinic in metropolitan Melbourne, Australia to a Skilled Parenting group or to a Perceptive Parenting group according to parent preference.⁹¹ Greater reductions in SDQ total difficulties were reported in the Skilled Parenting group [baseline mean: 24.18 (4.70), post-treatment mean: 20.77 (4.77)] than in the Perceptive Parenting group [baseline mean: 24.17 (4.85), post-treatment mean: 23.44 (7.54)] over 8 weeks of active treatment.

One study sequentially assigned parents of 3 to 9 year old children (73% male) referred with disruptive behaviors to an outpatient clinic in Ohio to receive the Helping the Noncompliant Child parent intervention or to treatment as usual.⁹⁰ Although children in both groups improved on parent-reported measures of child disruptive behaviors, change from baseline to post-treatment in parent-reported child disruptive behaviors as measured by the CBCL Total Problems subscale did not differ significantly between children referred to the Helping the Noncompliant Child intervention group [baseline mean: 68.7 (8.8), post-treatment mean: 64.3 (11.1)] and treatment as usual group [baseline mean: 68.7 (8.4), post-treatment mean: 64.3 (11.3)] or on other parent-reported measures of other constructs.

One RCT randomized 2 to 10 year old children (85% boys) referred for disruptive behaviors to an outpatient clinic in the United Kingdom to receive intensive outpatient treatment or standard treatment.¹⁴⁷ The intensive treatment differed from the standard program primarily by its inclusion of 3 5-hour sessions that included individual units and videotaped recording of parent-child interactions in order to give feedback to parents (average service contact 25 hours in 11 visits over 24 weeks) to the standard outpatient treatment (average service contact 7 hours in 6 visits over 24 weeks). Child disruptive behavior was measured by parent-reported CBCL Externalizing subscale. Parent-reported child disruptive behaviors as measured by the CBCL mean Externalizing T-score for the standard treatment group [baseline mean: 75.3 (5.9), post-treatment: 67.0 (9.23)] and intensive group [baseline mean: 74.2 (9.28), post-treatment mean: 63.9 (11.1)] both decreased from baseline to post-treatment and statistical models showed a main effect for time but no group-by-time interaction. Importantly, only the intensive treatment group had a mean score below the clinical cut-off at post-treatment. A companion paper reporting 4-year followup reported that the intensive treatment group's mean CBCL Externalizing scores remained below the clinical cutoff and that the standard treatment group meaning CBCL scores had worsened such that improvement from baseline was no longer evident at 4-year followup.¹⁶⁷

Incredible Years (IY)

Four studies examined IY components delivered in combination with each other (3 studies of IY-PT + IY-CT; 1 study of IY-PT + IY-CT + IY-TT) for school-age children (Table 18).^{105,110,126,131}

Table 18

Summary of behavior outcomes for studies of IY interventions in school-age children.

One RCT randomized 4 to 8 year old children (80% boys) referred to two child psychiatry outpatient clinics in Norway due to oppositional or conduct problems to IY-PT, IY-PT plus IY-CT, or to a waitlist control group.¹²⁶ Mother-reported child disruptive behaviors as measured by the ECBI Intensity scale were significantly reduced for IY-PT [baseline mean: 157.1 (24.2), post-treatment mean: 116.5 (27.0)] as compared to the waitlist control group [baseline mean: 159.7 (23.1), post-treatment mean: 137.3 (28.6)] but no significant difference between the IY-PT plus IY-CT [baseline mean: 156.5 (22.0), post-treatment mean: 121.8 (31.9)] and waitlist control group. Mother-reported aggressive behavior as measured by the CBCL Aggression subscale was significantly reduced for the IY-PT [baseline mean: 18.8 (6.8), post-treatment mean: 110 (7.0)] and IY-PT + IY-CT [baseline mean: 21.7 (7.0), post-treatment mean: 13.7 (8.6)] as compared to the waitlist control group [baseline mean: 20.0 (7.7), post-treatment mean: 17.2 (8.2)]. No significant between-group difference was reported for mother-reported ECBI-P. Generally, father-reported child disruptive behaviors correlated strongly with mother-reports.

One RCT included 4 to 8 year old children (80% boys) referred for treatment to two child psychiatric outpatient clinics in Norway by parents due to disruptive behaviors.¹³¹ Children and their parents were randomized to receive IY-PT, IY-PT + IY-CT, or to a waitlist control condition. The PT groups lasted 12-14 weeks. The CT sessions took place over 18 weeks. Results indicate a significant main effect for group on teacher-reported aggression as measured by the Preschool Behavior Questionnaire (PBQ) for children in day care and the aggression subscale of the TRF for children in school from baseline to post-treatment, co-varying baseline scores [PT + CT baseline mean: 3.0 (1.6), PT + CT post-treatment mean: 1.8 (1.5); PT-only baseline mean: 2.7 (1.7), PT-only post-treatment mean: 2.5 (1.4); WLC baseline mean: 3.2 (1.6), WLC post-treatment mean: 3.1 (1.6)]. Teacher-reported child disruptive behavior was significantly reduced in the PT + CT group in comparison to the PT-only (p<0.05) and waitlist control (p<0.01) groups but the PT-only and waitlist control groups did not significantly differ.

One RCT randomly assigned 4 to 8 year old children (75% boys) referred to an outpatient university research clinic in the United States with conduct problems to receive IY-CT, IY-PT, combined IY-CT + IY-PT, or to a waitlist control condition.¹¹⁰ Families were assessed at baseline and 8 months post-baseline (2 months after 6 months of active treatment), and 1-year post-treatment (e.g., 1.5 years post-baseline). At 1-year post-treatment followup, there were significant effects for time for all three active treatment groups on all mother and father-reports of child disruptive behaviors (CBCL Total Behavior problems score, ECBI Intensity score, and PDR score), and child social problem solving via WALLY but no significant group-by-time interactions for any of these variables. Considering all effects together, the IY-CT + IY-PT was superior to IY-CT in that it had an effect on parenting and child behaviors, and was superior to IY-PT in that it had an impact on child social problem solving.

One RCT examined the effect of IY-PT + IY-CT +IY-TT when delivered together in comparison to other combinations of IY components, to individual IY components, and to a waitlist control condition in 4 to 8 year old children (90% boys) with ODD who were referred by their families to an outpatient university research clinic in the United States.¹⁰⁵ Children were randomly assigned to one of the following treatment conditions: IY-PT; IY-PT + IY-TT; IY-CT; IY-CT + IY-TT; IY-PT + IY-CT + IY-TT; or a waitlist control. Although the study mainly reports results from composite measures made up of a number of previously validated measures (composite measures) because they are not themselves validated measures, are excluded from this report, it also reports the percentage of children showing clinically significant improvements at 6 months (post-treatment) and 1-year followup. At 1-year followup (e.g., the last followup), the treatment arms with the highest proportion of children showing clinically significant improvements on mother-reported ECBI-I scores were the IY-PT +IY-TT (84.6%) and IY-CT + IY-TT (81.3%) groups, but the only significant contrasts were between the IY-PT + IY-TT and IY-PT groups (with the combined treatment showing greater change) and the IY-CT group showing more improvement than the IY-PT group. It should also be noted that on teacher reported aggression via the TASB that the IY-CT group was more likely to have shown clinical improvement than the IY-PT + IY-CT + IY-TT group.

Coping Power Program

Two studies^128,134 assessed the Coping Power Program (Table 19).

Table 19

Summary of behavior outcomes for studies of Coping Power Program for school-age children.

One RCT assigned children ages 8 to13 years referred for disruptive behaviors to one of four child psychiatric outpatient clinics or three mental health centers in the Netherlands over a 3-year period to receive either the Coping Power Program (CPP) or treatment as usual.^128,168 Significant group-by-time interactions were reported only for PDR overt aggression subscale [CPP baseline mean: 2.90 (1.51), CPP post-treatment mean: 1.90 (1.38); treatment as usual baseline mean: 2.46 (1.53); treatment as usual post-treatment mean: 2.05 (1.43)], but not for the other parent-reported measures of child disruptive behavior (e.g., PDR oppositional behavior subscale, CBCL Externalizing Behavior subscale) or for the teacher-reported measure of child disruptive behavior via the TRF externalizing behavior subscale. At 5-year followup, there were no significant differences between the CPP or TAU groups on the NYS Delinquency Scale [CPP mean: 1.2 (1.5); TAU mean: 1.5 (1.5)], but children in the CPP group did report being less likely than children in the TAU group to smoke cigarettes in the past month (CPP % smoked in the last month = 17; TAU % smoked in the last month = 42) and lifetime use of marijuana (CPP % with lifetime marijuana use = 13; TAU % with lifetime marijuana use = 35).¹⁶⁸

A second RCT randomly assigned 278 children from 8 to 13 years of age in Puerto Rico who met DSM-IV-TR criteria for a disruptive behavior disorder to receive a culturally sensitive cognitive behavioral intervention (n = 174) or to a waitlist control group (n = 104).¹³⁴ Behavioral outcomes were measured with the Irritability / Hostility subscale of the Bauermeister School Behavior Inventory. Although boys and girls in the treatment group demonstrated more improvement on this subscale score over 12 weeks of than did children in the control group, these differences were not statistically significantly different.

Modular

Two studies^97,101 (each including multiple papers) examined the effects of a modular treatment for children with ODD or CD (Table 20). The modular treatment included seven components: (1) child CBT/skills training, (2) child medication for ADHD, (3) parent management training, (4) parent-child / family therapy, (5) school programming/teacher consultation, (6) peer relations/community activities development, and (7) case/crisis management.

Table 20

Summary of behavior outcomes for studies of modular intervention in school-age children.

One RCT included children aged 6 to11 years (85% boys) referred for disruptive behavior disorders to program sites associated with a university medical center in the United States.¹⁰¹ Children and families were randomly assigned to receive the modular treatment either in the community or in an outpatient research clinic setting. Healthy controls were included to provide norms for self-report questionnaires. Results suggest significant improvement in both groups from baseline to post-treatment (6 months) on measures of child disruptive behaviors including the CBCL Externalizing subscale, IOWA Conners Rating Scale oppositional defiant subscale, Self-Report of Antisocial Behavior (SRAB), TRF externalizing behavior, and Child and Adolescent Functional Assessment Scale (CAFAS). There were not group-by-time interactions from baseline to post-treatment or at 3-year followup for any measures indicating that the modular treatment can be successfully implemented in a research clinic or community based setting.

One RCT examines the effectiveness of the same modular treatment adapted for implementation by nurses in primary care settings.⁹⁷ To examine this, children aged 6 to 11 years (65% boys) were enrolled based on parent concerns about disruptive behaviors and scores on the Pediatric Symptoms Checklist (or PSC-17) above the clinical cutoff for externalizing behavior problems to either the nurse-administered modular care (PONI) or to enhanced usual care (EUC). From baseline to 1-year followup, significant group-by-time interactions were seen on the Individualized Goal Achievement Rating (IGAR) average [PONI baseline mean: 1.0 (0.0), post-treatment mean: 2.8 (0.8); EUC baseline mean: 1.0 (0.0), EUC post-treatment mean: 2.6 (0.8)] and Child Health and Illness Profile (CHIP) total score [PONI baseline mean: 47.3 (5.9), PONI post-treatment mean: 49.5 (5.9); EUC baseline mean: 48.7 (6.0), EUC post-treatment mean: 48.9 (6.1)], but not on the parent-reported PSC-17 externalizing score, parent-reported SDQ total score, or teacher-reported SDQ total score even though both groups reported change over time for almost all of these measures.

SNAP Under 12 Outreach Project (SNAP Under 12 ORP)

Two studies evaluated the SNAP Under 12 in Canada (Table 21).^89,137

Table 21

Summary of behavior outcomes for studies of the SNAP Under 12 ORP intervention in school-age children.

One RCT included children (mean age approximately 9 years; approximately 75% boys) who had police contact within 6 months of referral and/or a T-score on the CBCL Delinquency subscale indicating behavior problems more serious than 98 percent of same-age and same-sex peers.¹³⁷ Children were randomized to receive the SNAP Under 12 12-week outpatient program or to a waitlist control group that participated in a recreation group called the Cool Runner's Club. From baseline to 3 months, children in the SNAP Under 12 group experienced significantly greater declines in parent-reported child disruptive behaviors as measured by the CBCL Delinquency [baseline mean: 8.9, 3-month mean: 4.9, SDs not given] and aggression [baseline mean: 18.8, 3-month mean: 15.5, SDs not given] subscale mean scores than did children referred to the waitlist control group [delinquency subscale baseline mean: 8.9, delinquency subscale 3-month mean: 8.4; aggression subscale baseline mean: 19.4, aggression subscale 3-month mean: 19.0, SDs not given]. At the 3-month point, the two groups switched treatments and from 3 months to 18 months, the children originally referred to SNAP Under 12 continued to make progress and the children originally referred to the waitlist control (who were now receiving SNAP Under 12) also showed improvement (although they never caught up to the other group) on the same measures.¹³⁷

In a second study, investigators recruited boys ages 6 to 11 years from the community to participate in SNAP Under 12 using similar inclusion and exclusion criteria as described in the previous study but allocated children to SNAP Under 12 or the waitlist control recreation group on a first-come, first-served basis rather than being randomized.⁸⁹ In comparison to children initially referred to the waitlist control condition, children initially referred to the SNAP Under 12 showed significantly more improvement on parent-reported child disruptive behaviors as measured by the CBCL Rulebreaking, Aggressive, Conduct Problems, and Total Problems subscales, but not on the CBCL Competence subscale or TRF outcomes.

Other Multicomponent Interventions

Seven studies, each evaluating a different multicomponent intervention, were also identified (Table 22).^{92,96,100,103,108,123,165}

Table 22

Summary of behavior outcomes for studies of other interventions in school-age children.

One RCT evaluated the effectiveness of a CBT as compared to an educational fire safety intervention or a home visit from a firefighter for fire-setting behavior in 54 boys aged 5 to 13 years and referred due to documented fire-setting behavior by the City of Pittsburgh Bureau of Fire, direct parental solicitation, or a mental health practitioner.¹⁰⁸ Children in each of the three intervention groups showed significant improvement on measures of fire involvement, interest and risk, but CBT and fire safety intervention were not more effective at reducing fire-setting behaviors even though the group-by-time interaction approached statistical significance (p<0.06).¹⁰⁸

One RCT assigned children between the ages of 4 and12 years (68% boys) clinically referred to an outpatient mental health clinic specializing in the treatment of disruptive behavior disorders at a university teaching hospital and meeting criteria for ODD to receive Collaborative Problem Solving (CPS) or parent training based on Barkley's (1997) 10-week behavior management program.¹⁰³ The primary measure of child disruptive behavior was the parent-rated Oppositional Defiant Disorder Rating Scale (ODDRS). On this measure of parent-reported child disruptive behaviors, there was a significant change from baseline to post-treatment and from baseline to 4-month followup for children in the CPS group, but the group-by-time interactions from baseline to post-treatment and from baseline to 4-month followup were not significant (means for both groups at each time point are not given). The group-by-time interactions on the Parent-Child Relationship Inventory (PCRI), a measure of one of this reviews functional outcomes, was also not significant.

One RCT assigned children between 8 and11 years (73% boys) with mood disorders and their families to receive either treatment as usual plus immediate treatment in the multifamily psycho-education program (MF-PEP) or treatment as usual plus waitlist control.⁹⁶ Disruptive behaviors were measured with the Children's Interview of Psychiatric Syndromes (ChIPS) and Parent Form (P-ChIPS). Although MF-PEP was associated with a significant decrease in ODD symptoms from baseline to 12 months followup, there was no significant difference between the MF-PEP [baseline mean: 5.7 (2.1), 12-month followup: 4.5 (2.6)] and the waitlist control groups [baseline mean: 5.4 (2.6), 12-month followup: 4.9 (2.7)] on ODD symptoms. There was also no difference in CD symptoms from baseline to 12-month followup.

One RCT assigned parents of children with a mean just over 5 years of age (71% boys) who were screened for disruptive behaviors with the SDQ in schools in London to either receive the Supporting Parents on Kids Education in Schools (SPOKES) intervention or to receive access to a telephone hotline designed to help parents access treatment as usual in the community over 28 weeks of active treatment.¹²³ In comparison to children in the control group, children receiving SPOKES had significant reductions in child antisocial behavior as measured by parent interview [SPOKES baseline mean: 1.15 (0.44), SPOKES post-treatment mean: 0.91 (0.36); treatment as usual baseline mean: 1.12 (0.49), treatment as usual post-treatment mean: 1.13 (0.49)], parent-reported child disruptive behavior as measured by the ECBI Intensity subscale) [SPOKES baseline mean: 119.1 (31.6), SPOKES post-treatment mean: 103.9 (27.3); treatment as usual baseline mean: 115.9 (27.0), treatment as usual post-treatment mean: 113.2 (31.3)], but little difference in teacher-reported oppositional symptoms as measured by a DSM-IV questionnaire items.

An additional RCT examining Project Support and was conducted in the United States. Authors examined the effectiveness of Project Support, a family intervention specifically designed to reduce disruptive behaviors in the children of women at a domestic violence shelter.¹⁰⁰ The intervention provides mothers with child behavior management skills and instrumental and emotional support.¹⁰⁰ Although therapists worked primarily alone with mothers, children were regularly included in sessions so that mothers' skill using the new techniques could be evaluated and additional skill building activities could be tailored according to the child's response to them. Child conduct problems were measured by two maternal self-report measures (CBCL Externalizing and ECBI Intensity). Mean CBCL Externalizing scale scores decreased from 67.9 to 57.4 pre- to post-treatment for the Project Support group and from 65.9 to 61.6 for the treatment as usual control group (Cohen's d=0.66) and from 142.1 to 102.5 and 129.8 to 102.7 on the ECBI Intensity scale (Cohen's d=0.17), respectively.

One moderate risk of bias, prospective cohort study⁹² compared children sequentially assigned to a multimodal treatment program (MTP), which includes once a week sessions for 1 year of individual and group support for children and individual parent training, or treatment as usual (TAU). The study sample consisted of 135 youth with a mean age of 12.0 (2.5) years. Mean CBCL Externalizing scores decreased in the MTP group from 69.73 (7.43) at baseline to 65.58 (7.34) at end of treatment and from 71.49 (7.25) at baseline to 68.58 (7.62) at end of treatment in the treatment as usual group. Mean CBCL Aggressive Behavior scores decreased in the MTP group from 71.67 (9.03) at baseline to 66.81 (8.52) at end of treatment and from 74.06 (9.77) at baseline to 71.17 (10.00) at end of treatment in the TAU group. The mean CBCL Delinquent Behavior scores decreased in the MTP group from 66.03 (8.07) at baseline to 63.42 (7.51) at end of treatment and in the TAU group from 67.90 (8.31) at baseline to 65.20 (7.92) at end of treatment. This group-by-time interaction was not statistically significant.

Finally, one high risk of bias RCT¹⁶⁵ compared children assigned to a reciprocal skills training (RST), a family-based treatment, against children assigned to a waitlist control group. The study sample consisted of 57 children with a mean age of 8.47 (1.6) years. The intervention group consisted of children referred from a clinic setting and from a pre-treatment hospital setting. Because studies of inpatient hospital settings are excluded from this review, only results for the clinic setting are reported here. On the parent-reported CBCL Externalizing scale, mean scores for the clinic-referred RST group decreased from 67.4 (7.0) at baseline to 59.8 (11.5) end of treatment and from 70.0 (5.8) to 74.0 (5.0) for the waitlist control group. The group-by-time interaction effect was statistically significant. In addition, the percentage of children who no longer met DSM-IV criteria for oppositional defiant disorder was significantly reduced in the clinic-referred RST group than in the waitlist control group (72.2% vs. 30%, p<0.01).

Family Therapy

Six studies examined the impact of family therapy interventions on disruptive behaviors and other related outcomes (Table 26).^106,142-144 Three of these studies were conducted in Germany^142-144 and three were conducted in the United States.^106,115,116 Two of the studies measured disruptive behaviors using the self-reported Adolescent Risk Taking Behavior Scale (ARBS),^142,143 while another study used the conduct disorder and socialized aggression subscales of the parent-reported Revised Behavior Problem Checklist (RBPC) behavior problem scale.¹⁰⁶ Three studies also measured levels of anger and anger expression using the self-report State-Trait Anger Expression Inventory (STAXI)^142-144 Three studies also included outcomes related to health related quality of life,^142-144 two included interpersonal functioning outcomes^142,143 and one measured the impact of the intervention on family functioning.¹⁰⁶

Table 26

Summary of studies of multicomponent interventions (family therapy) for teenage children.

Three of the four studies examined the delivery of BSFT and its impact on disruptive behavior problems and related outcomes.^106,142,144 In each of these studies the intervention group receiving BSFT was compared to a control group intervention in which the participants received either group therapy¹⁰⁶ or another family-based intervention.^142,144 Each of these studies examined the use of BSFT with a specific population. One study¹⁰⁶ compared the effectiveness of BSFT for a primarily male Hispanic adolescent population (n = 126, mean age: 15.6 years; 75% male) with a general group therapy based intervention. Participants were included based on parental or school complaints of externalizing behavior problems. Compared to the control group, participants receiving the BSFT intervention displayed a significantly greater reduction in behavior problems as measured by the Revised Behavior Problem Checklist (RBPC). Compared to 11 percent of clinically significant improvement in the control group (p=NS), 43 percent of the BSFT group showed reliable improvement in Conduct Disorder measures (p<0.001). Similarly, on the Socialized Aggression scale, 36 percent of BSFT recipients showed reliable improvement (p<0.001) compared to 11 percent of the control population (ns). The treatment group also reported significant reductions in substance use and increased improvements in family functioning as compared to the control group.

Two studies evaluated the impact of BSFT on bullying behaviors: one¹⁴² in a population of adolescent females and the other¹⁴⁴ with adolescent males. The first study compared the effectiveness of BSFT on bullying behavior in a group of 15 year old girls [n = 40, mean age: 15.5 (0.5) years] who had shown direct verbal and/or physical bullying behavior for at least six months to a placebo intervention. The study assessed risk-taking behaviors using the Adolescent Risk Taking Behavior Scale (ARBS), which consisted of seven behavior scales: drug use, smoking, binge drinking, excessive media use, having sex without a condom, having sex while using drugs and alcohol and sexual disinhibition. The study found that girls receiving the BSFT showed significantly greater reductions in adolescent risk taking behavior than those receiving the placebo treatment at both the end of treatment (ARBS score mean difference between groups: −9.3, p<0.001) and after a one year followup assessment (ARBS score mean difference between groups: −8.2, p<0.001). BSFT also led to significant improvements in interpersonal relationships as measured by the Inventory of Interpersonal Problems (IIP-D), as well as reducing levels of anger (STAXI) and increasing health related quality of life (SF-36) as compared to the placebo intervention. These results were reported to have remained relatively stable at one-year followup.

One study¹⁴³ examined the effectiveness of family therapy as a monotherapy for reducing disruptive behaviors and anger compared to a placebo intervention. This study utilized an integrative family therapy model that integrated elements from family systems theory, psychodynamic-oriented therapy, gestalt therapy, and behavioral therapy. Interventions were focused around communication, family rules and each family member's role in the existing problematic family system presentation. Forty-four male adolescents [mean age: 15.2 (0.5)] displaying bullying behavior participated in the study, half randomly assigned (n = 22) to receive a family therapy program for 6 months and the other half assigned (n = 22) to the placebo control group for the same length of time Consistent with the results from the female [mean age: 15.5 (0.5)] cohort study,¹⁴² this study reported significantly greater reductions in adolescent risky behaviors on all scales of the ARBS (end-of-treatment ARBS score mean difference: −6.3, p<0.001; followup ARBS score mean difference: −3.1, p<0.001) and significant reductions in anger levels on nearly all of the scales measured by the STAXI. Additional reported outcomes included significant improvements in interpersonal relationships, as measured by six of the eight scales on the IIP-D, and significant improvement in health related quality of life (SF-36) as compared to the placebo control group.

One high risk of bias RCT¹¹⁵ compared children assigned to family-behavioral therapy with children assigned to individual cognitive therapy. The study sample consisted of 56 children with a mean age 15.4 (1.3) years. On the parent-reported CBCL Delinquency scale children assigned to the family-behavioural therapy group experienced greater reductions [baseline mean score: 74.44 (6.70); end of treatment mean score: 63.55 (9.10)] than did children assigned to the individual cognitive therapy group [baseline mean score: 77.40 (8.45); end of treatment mean score: 66.67 (12.11)]. These differences were maintained at 6-month followup. Similar findings were also evident via the parent reported ECBI Problem scale (family behavioural therapy group baseline mean score: 17.86 (8.52); family behavioural therapy group end of treatment mean score: 8.58 (9.09); individual cognitive therapy group baseline mean score: 21.52 (6.12), individual cognitive therapy group end of treatment mean score: 11.95 (9.46)], and ECBI Intensity scale (family behavioural therapy group baseline mean score: 133.55 (38.26); family behavioural therapy group end of treatment mean score: 90.78 (36.37); individual cognitive therapy group baseline mean score: 145.93 (35.58); individual cognitive therapy group end of treatment mean score: 110.35(45.92)].

One moderate risk of bias RCT¹¹⁶ compared Parenting with Limits and Love (PLL), a 6-week group therapy program integrating principles of a structural family therapy approach, against a control group receiving TAU probation services including counseling, community schools, and/or community service. The study sample included 38 teenagers [mean age: 15] (57% boys) who had been referred for criminal offenses. Disruptive behaviors were assessed via the parent-reported CBCL. Mean scores in the intervention group showed greater decrease than in the control group on the CBCL Externalizing subscale (intervention group baseline mean score: 64.07 (15.80), intervention group end of treatment mean score: 56.57 (11.21); control group baseline mean score: 73.08 (9.54), control group end of treatment mean score: 71.83 (10.11)], aggressive behaviors scale (intervention group baseline mean score: 67.43 (12.77), intervention group end of treatment mean score: 58.14 (6.78); control group baseline mean score: 70.83 (14.22), control group end of treatment mean score: 71.67 (13.01)], and rule-breaking behaviors scale (intervention group baseline mean score: 67.29 (10.94), intervention group end of treatment mean score: 60.07 (8.07); control group baseline mean score: 75.33 (7.30), control group end of treatment mean score: 69.33 (9.44)]. The group-by-time interaction for each of these measures was statistically significant.

Multisystemic Therapy

Of the five studies that examined MST, two were conducted in the United States,^94,111 and one each in the Netherlands,¹³⁶ the United Kingdom,¹²⁰ and Sweden.¹²⁴ All five of these studies were RCTs (1 high, 2 moderate, and 2 low risk of bias). Overall, the treatment effects were positive, with only one study¹²⁴ not demonstrating significance (Table 27).

Table 27

Summary of studies of multicomponent interventions (MST) for teenage children.

One study^94,172 conducted in the United States randomly assigned 164 adolescents (83% male) between the ages of 11 and 18 years in one school system's self-contained behavior intervention classrooms to receive MST or treatment as usual. Treatment as usual included behavior management interventions and support provided as part of the classroom structure. At 18-month followup, parent-reported CBCL Externalizing mean scores for the MST group [baseline mean: 25.90 (10.63); end of active treatment mean: 18.20 (10.82); change: 30% reduction] decreased significantly more from baseline to end-of-treatment than those of the control group [baseline mean: 23.40 (9.61); end of active treatment mean: 19.19 (10.36); change: 18% reduction]. The outcomes from the YSR assessment showed similar results, with the MST group [baseline mean: 17.63 (9.03); end of active treatment mean: 13.87 (8.53); change: 21% reduction] showing greater effects than the control group [baseline mean: 17.00 (7.97); end of active treatment mean: 14.22 (7.72); change: 16% reduction]. No significant effect was found based on the TRF of externalizing behaviors or arrest data.

One pretest-posttest control group design (moderate risk of bias)¹¹¹ conducted in the United States compared the effects of MST to individual therapy (IT) on criminal behavior and violent offenses among a group of high-risk juvenile offenders (n = 176, 67% male). Ninety-two participants [mean age: 14.8 (1.5)] were randomly assigned to receive MST, with 77 completing both pre- and post-treatment assessments and receiving an average of 23.9 (8.2) hours of treatment. Out of the 84 participants initially assigned to the IT control group, 63 completed both assessments and received an average of 28.6 (9.8) hours of treatment.

This study demonstrated MST to be significantly more effective than individual therapy based on several outcome measures. From pre-treatment to post-treatment, both mothers and fathers from the MST group showed significant decreases in psychiatric symptomology as measured by the SCL-90-R [mother mean baseline score: 0.12 (1.02); mother mean post-treatment score: −0.15 (0.97); father mean baseline score: −0.06 (0.90); father mean post-treatment score: −0.07 (0.77)]. Their IT counterparts did not show similar reductions in psychiatric symptomatology for either of the parents [mother mean baseline score: 0.04 (1.17); mother mean post-treatment score: 0.20 (1.26); father mean baseline score: 0.06 (1.05); father mean post-treatment score: 0.19 (1.09)]. The study also showed a significant interaction effect for mothers' reports of adolescent disruptive behaviors as measured by the Revised Behavior Problem Checklist (RBPC), with mothers in the MST group reporting a decrease in adolescent behavior problems and mothers of youths in the IT group reporting an increase in behavior problems. Additionally, adolescents in the MST group showed significant positive change in family functioning and cohesion (FACES-II), lower re-arrest rates, and less serious offenses when rearrested. The pattern of lower frequency and decreased seriousness of crimes emerged in both the analysis of the entire sample as well as when analyzing only those that completed treatment.

Three studies^120,124,136 examined the effectiveness of MST to treatment as usual in more socialized systems offering comprehensive management of disruptive behavior problems (i.e., United Kingdom, Sweden, and The Netherlands). One low risk of bias study¹²⁰ conducted in the United Kingdom compared MST to outcomes for youth working with a Youth Offending Team (YOT). YOTs, like MST, provide a multicomponent intervention that is led by a social worker working with additional team members, such as therapists and probation officers. This study examined the impact of MST versus YOT, or usual services, on offending behavior based on police records (primary outcomes), as well as parent and youth rated reports of disruptive and delinquent behaviors as measured by the CBCL and YSR (secondary outcomes). A group of 108 adolescents between 13 and 17 years of age were allocated to receive either MST [n = 56; mean age: 182.7 (12.3) months; 91% male] or YOT [n = 52; mean age: 180.6 (12.9) months; 90% male]. Based on data derived from police computer records, youth who participated in MST had significantly less nonviolent offending by the end of the followup period (18 months post treatment end). There were no significant differences with regard to violent offending given the low number of youth with violent offense records.

In regards to secondary outcome measures, assessments from baseline to 6 months post-treatment indicated that, for internalizing and externalizing problems, there was no significant difference in disruptive behaviors between the two groups. However, the CBCL scales pertinent to the hypothesis each showed significant interactions favoring MST. For the aggression subscale, MST participants showed significantly more improvement [baseline mean: 69.4 (12.9); 6-month mean: 64.2 (11.4); change: 7.5% reduction] than the YOT group [baseline mean: 66.9 (11.6); 6-month mean: 65.9 (11.9); change: 1.5% reduction]. Similar results occurred with the delinquency subscale, with MST participants again showing significant reductions in [baseline mean: 73.4 (8.3); 6-month mean: 67.9 (8.6); change: 7.5% reduction] compared to the YOT group [baseline mean: 73.0 (7.9); 6-month mean: 70.9 (8.5); change: 2.9% reduction]. Analysis of rates of change also indicated moderate effect sizes in the MST group (aggression effect size: 0.42; delinquency effect size: 0.64) and smaller effect sizes for the YOT group (aggression effect size: 0.09; delinquency effect size: 0.25). While the parent-reported outcomes suggested improvement in disruptive behaviors in the MST group, none of the scales from the YSR yielded significant interactions. Data regarding longer-term follow up of rate of disruptive behavior were not available.

One multicenter low risk of bias study¹²⁴ examined the effectiveness of MST to treatment as usual in Sweden. Treatment as usual for court-referred youth in Sweden includes referral for social service supports, which work to identify treatment needs. Treatment in the control group was varied and was primarily represented by individual therapy, family therapy, mentoring, or no services. A group of youths between the ages of 12 and 17 fulfilling diagnostic criteria for conduct disorder [n = 156; mean age: 15.0 (1.4) years; 61% male] were randomized to either the treatment (n = 79) or control group (n = 77). Mean enrollment in MST lasted 145.8 (51.6) days. Disruptive and delinquent behavior was assessed by both caregiver and adolescent ratings through the CBCL and YSR, respectively. Additionally, the study looked at delinquency through the Self-Report Delinquency Scale (SRD), substance use measures through multiple self-reporting methods (i.e. AUDIT/DUDIT), and relationships and social competence (i.e. Pittsburgh Youth Study, SCPQ, Social Skills Rating System, school attendance). Pre- to posttest measurements did not demonstrate any significant differences in the MST intervention compared to treatment as usual as measured by the CBCL and YSR measures, nor on the SOC scale. Both groups showed decreased disruptive and delinquent behavior, improvement in social skills and better family relations.

One RCT (high risk of bias)¹³⁶ examined the effectiveness of MST compared to treatment as usual in The Netherlands. In The Netherlands, treatment as usual relies more frequently on in-home services, but also includes individual treatment, some combination of both or no services. The study included 256 adolescents [mean age: 16.02 (1.31) years; 73% male] randomly allocated to either MST or treatment as usual interventions. Researchers used the CBCL Aggression and Delinquency subscales to assess externalizing behaviors and delinquency. Parents also filled out several symptom scales from the Disruptive Behaviors Disorder rating scales. Adolescents self-reported using YSR and SRD assessments. According to both parent and youth self-reports, MST was significantly more effective at reducing externalizing behavior problems [CBCL baseline mean: 23.32 (12.60); CBCL end-of-treatment mean: 17.64 (11.57); change: 24% reduction] than treatment as usual [CBCL baseline mean: 22.55 (12.95); CBCL end-of-treatment mean: 19.25 (10.56); change: 15% reduction]. The YSR showed similar results, with a 16 percent reduction in the MST group and only a 3 percent reduction in the treatment as usual group. MST was also more effective at decreasing ODD and CD, as compared to treatment as usual. With regard to self-report of delinquent behaviors, MST demonstrated significant reductions for property offenses, but no significant effect was found for violent offending. Interestingly, further analysis of other secondary outcomes- such as parent and adolescent cognitions, parenting behavior and peer relationships- and demographic variables yielded unexpected results. While MST was equally effective across ages and ethnicity, the intervention showed larger effects for adolescent cognitions for boys than for girls. At 6-month post-treatment follow-up, there was evidence of sustained effects of MST in comparison to TAU with maintenance of statistically significant reductions in externalizing problems, ODD, and CD, but the number of re-arrests and time to re-arrest did not differ between the groups.¹⁷¹

Other Multicomponent Interventions

One RCT (moderate risk of bias)¹⁴⁶ examined the effect of a semi-structured bibliotherapy intervention aimed to decrease aggressive behavior in youth (Table 28). The study was conducted in the Druze community in Israel, which is a closed society living in generally segregated cities or villages. Seventy-five children (77% male) were randomly and equally assigned to one of three conditions: child treatment only, mother plus child treatment and no treatment at all. The additional parent group was aimed at increasing parent's understanding of their child's aggressive behavior. Researchers measured aggression using a reduced version of the aggression and delinquency subscales of the CBCL questionnaire as reported by parents (CBCL), the adolescents (YSR), and their teachers (TRF). Another parent report, Coping with Children's Negative Emotions Scale (CCNS), was also used. Both treatment groups, the child group and the parent/child combination group, were more effective at reducing aggressive behavior than no treatment at all. However, the combined intervention was not significantly more effective at reducing disruptive behaviors than the child training only intervention. While obtained means demonstrated a greater decrease in aggressive behavior of the combined treatment intervention, significance was only found with the self-report measure, not with the parent or teacher report. Thus, the researchers' hypothesis of enhanced outcomes with the additional parent component was only partially supported.

Table 28

Summary of studies of multicomponent interventions (other) for teenage children.

A nonrandomized cohort study (high risk of bias)⁸⁶ conducted in The Netherlands did not report positive treatment effect for disruptive behavior problems (Table 28). The study compared the effectiveness of treatments being offered in a forensic youth outpatient clinic at reducing recidivism. Treatments included functional family therapy (FFT) (n = 55), individual CBT (n = 87), and CBT combined with parent training (n = 50). In addition to these treatments, some youths also participated in Aggression Replacement Training (ART) (n = 27). It should be noted that both FFT and ART were implemented as trial versions and most implementing therapists had not been formally trained in administering these interventions. The official records of the 192 adolescents completing treatment in the outpatient clinic (mean age: 17.0 years; 85% male) were analyzed retrospectively, with occurrences of recidivism serving as the primary outcome measure. The study found no significant differences in 2-year total or violent recidivism rates between the different treatment interventions. However, researchers did find a higher recidivism rate for those youth who had additionally participated in ART (54% recidivism compared to 30% for non-ART juveniles), even after controlling for the type of offense committed (i.e. violent). There was also no significance found in recidivism between the treatment groups as compared to youth who dropped out of treatment (n = 42). The study found a significant interaction between moderating variables regarding patient characteristics (ethnicity), intensity and frequency of treatment, and the therapist conducting the training.

Valproic Acid Versus Placebo

One randomized, placebo-controlled study (moderate risk of bias)¹⁷⁸ measured the effect of valproic acid in reducing aggressive behavior in younger children, from ages 6 to 13 years, with 21 of 27 males, who had aggression persisting after a trial of stimulant medications. Thirty patients were randomized to add-on valproic acid or placebo adjunctive to stimulant medication for eight weeks. The study participants were boys (n = 21) and girls (n = 6) with a diagnosis of ADHD and a co-diagnosis of either ODD or CD. Enrollment occurred between 2004 and 2007 at two academic medical centers in the United States. The mean daily dose of children in the valproic acid group was 567 mg/day (mean serum level: 68.11 mg/liter) and the children assigned to the placebo group had a drug equivalent dose of 685 mg.

The primary outcome was scoring on the Retrospective-Modified Overt Aggression Scale (R-M OAS). Thirteen patients in the placebo group and 14 patients in the valproic acid group were analyzed due to withdrawal prior to first assessment. The scores on the R-M OAS dropped from 41.80 to 32.13 in the valproic acid group and from 53.31 to 35.77 in the placebo group, with no significant difference between groups.

Another placebo-controlled crossover RCT (moderate risk of bias)¹⁸⁵ included children and adolescents, ages 10 to 18, mostly male (16/20) with conduct disorder or oppositional defiant disorder. The study was conducted at an outpatient clinic at an academic medical center in the United States. A blinded assessor rated the modified OAS and the SCL-90 Anger Hostility items. Response was measured as greater than or equal to 70 percent decrease from baseline in the combined scores of these items. The final dose of valproic acid ranged from 750 to 1500 mg per day. In the first 6-week phase of the study, 10 patients were randomized to the valproic acid arm and eight patients responded. None of the 10 patients randomized to the placebo arm responded.

During the second 6-week phase of the study, the participants crossed over to the alternate intervention; six of seven nonresponders to placebo in the initial phase achieved response in the treatment phase. Of the eight who switched from the treatment group to placebo in phase 2, all of whom had responded in phase 1, six relapsed.

High Dose Versus Low Dose Valproic Acid

One moderate risk of bias randomized, placebo-controlled study (reported in two publications)^184,189 measured the effect of valproic acid on a group of adolescent male patients with a diagnosis of conduct disorder from a correctional facility in California. The trial was 7 weeks long with 6 weeks of active treatment. Data were analyzed from 58 completers, all of whom had at least one offense “against persons.” The study included a (blinded) clinician-reported CGI. In the high dose group (mode=1000 mg/day, n = 34), 53 percent were “very much or much improved,” 29 percent were “minimally improved” and 18 percent were “no change or minimally worse.” In the low dose group (mode=125 mg/day, n = 24) 8 percent were “very much or much improved,” 42 percent were “minimally improved” and 50 percent were “no change or minimally worse.” The second paper of this family¹⁸⁹ focused on the difference between treatment with high or low dose valproic acid on High Distress Conduct Disorder (HDCD) and Low Distress Conduct Disorder (LDCD). In the high dose group, 25 were identified as HDCD and nine with LDCD. Of those with HDCD on high dose valproic acid, 16 showed a response as measured by CGI-I (defined as improved, much improved or very much improved) and nine showed no response (defined as No Response). Of those with HDCD on low dose valproic acid, two were responders and 14 showed no response. Of those with LDCD on high dose valproic acid, two were responders and seven showed no response. Of those with LDCD on low dose valproic acid, none showed response and eight showed no response. Overall, valproic acid appeared more effective at high dose than low dose, and more effective in the HDCD group than the LDCD group.

Amphetamine Salts (Adderall)

One high risk of bias multicenter, randomized, double blind, placebo-controlled study examined the efficacy and safety of mixed amphetamine salts extended release for the treatment of children and adolescents with ODD.¹⁸² Children and adolescents with ODD (n = 308) were randomized 1:1:1:1 to receive active treatment with mixed amphetamine salts extended release 10 mg/day (n = 60), 20 mg/day (n = 58), 30 mg/day (n = 69), or 40 mg/day (n = 61) or placebo (n = 60) for 4 weeks with forced dose escalation after a washout period. Eligible participants were aged 6 to 17 years with ODD as defined by DSM-IV-TR. Patients with conduct disorder were excluded. The primary outcome was the ODD subscale of the SNAP-IV parent rating. Secondary outcomes include the ODD subscale of the SNAP-IV teacher ratings, ADHD subscales of the SNAP-IV parent and teacher ratings, the child health questionnaire parent form 50 (CHQ-PF50) and adverse events. A total of 244 patients (79.2%) had comorbid ADHD, however results were not presented among this subgroup. The mean baseline score for the ODD subscale of the SNAP-IV parent rating did not differ by treatment group. In the intention-to-treat population, statistically significant improvements in oppositional symptoms as measured by the parent-rated SNAP-IV ODD subscale were observed in the least squares mean difference (−0.43) for those in the higher dose (30 mg/day) group compared with the placebo group (p<0.005). Statistically significant improvements for the teacher rated ODD subscale of the SNAP-IV from baseline to endpoint was seen in the intention-to-treat populations who received mixed amphetamine salts extended release 10 mg/day (p=0.047), 20 mg/day (p=0.043), and 30 mg/day (p=0.003), compared to placebo group. Mixed amphetamine salts extended release was associated with improvement in quality of life outcomes, measured with the CHQ, including statistically significant improvements in behavior, physical and psychosocial summary for those in the mixed amphetamine salts extended release 30 and 40 mg/day groups compared to placebo; and for self-esteem in the mixed amphetamine salts extended release 40 mg/day group compared to placebo. When stratified by baseline symptoms in a post hoc reanalysis of the per protocol population, mean changes from baseline in the ODD subscore of the parent rated SNAP-IV was greater for the high baseline ODD severity group.

Methylphenidate (Ritalin)

The second stimulant study that we identified compared methylphenidate to placebo in an RCT including 83 children and adolescents with CD.¹⁸⁷ Participants received methylphenidate (n = 41) or placebo (n = 42) for 5 weeks with a maximum dose of 60 mg per day in two divided doses to evaluate symptoms of CD. Eligible participants were 6 to 15 years of age and met DSM-II criteria for CD, which were slightly modified; moderate to severe impairment rating by teacher or parents, and an IQ greater than 70. DSM-IV criteria were used to diagnose ADHD. Primary outcomes were parent and teacher ratings of CD symptoms based on the Conners Teacher Rating Scale, and subscales of the Quay revised behavior problem checklist, and global estimates of the severity of behavioral problems. Participants mean age was 10.2 (2.3) years in the methylphenidate group and 10.2 (2.5) years in the placebo group. All but two children had at least three symptoms of CD, consistent with DSM-IV criteria; 69 percent of the population also met DSM-IV criteria for ADHD, however, results were given for the entire sample and not by those with comorbid ADHD separately. Baseline teacher overall rating for conduct problems was 2.6 (0.7). Teacher rated overall conduct problems were significantly less for those children taking methylphenidate [1.3 (0.1)] compared to placebo [2.3 (0.1), p<0.001] and factor scores including aggression, conduct problems, and conduct disorder were significantly improved compared to placebo. Teacher ratings of ADHD symptoms were also significantly improved among those who received methylphenidate compared to placebo. Significant improvements on parent ratings of aggression, conduct problems, and conduct disorder were seen in the methylphenidate group compared to placebo. Socialized aggression showed no statistical improvement on either the parent or the teacher ratings. Among 47 elementary school-aged children, classroom observers' ratings showed significant improvements among methylphenidate compared to placebo groups with regards to global rating of conduct problem severity and aggression (Iowa scale).

Atomoxetine

A moderate risk of bias randomized, double-blind, placebo-controlled, three-arm, multicenter study was conducted in 20 sites in Germany to assess the efficacy of atomoxetine given once daily for 9 weeks (target dose: 1.2 mg/kg/day), using either fast or slow titration, for treating symptoms of ODD in children and adolescents with ADHD and comorbid ODD/CD.¹⁷⁶ Eligible participants were aged 6 to 17 years and met the DSM-IV-TR criteria for ADHD (any subtype) and DSM-IV criteria A through C for ODD; DSM-IV-TR criteria for CD was not an exclusion. Only outpatients were enrolled from primary and secondary sites. Participants were randomized to one of three arms: (1) atomoxetine 0.5 mg/kg/day for 7 days followed by the target dose of 1.2 mg/kg (atomoxetine fast titrating group, n = 60); (2) atomoxetine 0.5 mg/kg/day for 7 days, followed by 0.8 mg/kg/day for 7 days, followed by target does of 1.2 mg/kg/day (atomoxetine-slow titrating group, n = 61); or (3) placebo (n = 59) for nine weeks, after a 3- to 28-day screening and washout period. The primary outcome was the investigator-rated SNAP-IV ODD subscale score. Other outcomes included the SNAP-IV ADHD subscale score and adverse events. Baseline characteristics were comparable for the three groups [84% male, mean age: 11 (3) years]. Participants DBD comorbidity was 74 percent (n = 134) ODD, 24 percent (n = 44) CD, with one patient meeting criteria for DBD, not otherwise specified and one for adjustment disorder. Baseline mean overall SNAP-IV ODD scores were 15.5 (4.35). Using a mixed effects model for repeated measures, treatment with atomoxetine once daily for nine weeks, pooling fast and slow titration arms, significantly reduced ODD symptoms compared to placebo, as measured by the SNAP-IV ODD score, least square mean treatment group difference at week 9, atomoxetine-pooled minus placebo: −3.2 (95% CI: −5.0 to −1.5), effect size: −0.69, p<0.001. The decrease in ODD symptoms was significant for both the fast and slow titration groups, (least square mean, atomoxetine-fast 8.6 (95% CI: 7.2, 9.9), atomoxetine-slow 9.0 (95% CI: 7.7, 10.3) compared to placebo 12.0 (95% CI: 10.6, 13.5), p<0.001, effects size −0.74 and p=0.003, effect size −0.65, respectively). SNAP-IV ODD scores improved at least 30 percent in 48.3 percent of patients in the atomoxetine fast titration group compared with 55.7 percent in the atomoxetine slow titration group and 35.6 percent in the placebo group. There were no significant differences between the atomoxetine fast and atomoxetine slow titration groups. Atomoxetine significantly reduced ADHD symptoms compared to placebo at week 9 as measured by the SNAP-IV ADHD subscale score. Patients in the atomoxetine slow titration group stayed on treatment significantly longer than did patients in the placebo group (HR=3.57; 95% CI: 1.42 to 8.94, p=0.007). Study was sponsored by industry.

A second paper in the family of studies evaluated the outcome of quality of life in the same 9-week trial of atomoxetine (target dose 1.2 mg/kg/day) versus placebo.¹⁹⁰ Quality of life was measured using the parent rated KINDL-R questionnaire total scores and sub-scores on emotional well-bring, self- esteem, friends, family, school, and physical well- being, a validated instrument. Family burden of illness was measured using the parent rated FaBel questionnaire, a German version of the Impact on Family Scale. At baseline, the mean overall KINDL-R scores were 62.9 (12.78) and the mean overall FaBel score were 53.8 (12.89). Among those treated with atomoxetine, the KINDL-R total score increased significantly compared to those in the placebo group, (mean change: 2.6 vs. −1.6 points) ANCOVA LS-mean difference, atomoxetine pooled minus placebo: 5.0 (0.8, 9.3), effect size: 0.377, p=0.021), which was clinically relevant. There was no significant difference between the fast and slow titration groups in KINDL total or subscores. Quality of life subscores for emotional well- being, self-esteem, family, and friends increased significantly in patients treated with atomoxetine compared to placebo. There were no statistically significant differences in the KINDL-R school subscore; however the subscore on physical wellbeing was significantly worse for patients in the atomoxetine group compared to placebo. Authors felt the physical wellbeing subscore differences may be related to common treatment adverse effects. No significant treatment effects were seen on family burden, as measured by the FaBel total score. However, the FaBel impact on siblings subscore improved significantly more in the atomoxetine group compared to placebo.

Finally, a multicenter, double blind, placebo-controlled trial (high risk of bias) conducted in Italy evaluated the efficacy of atomoxetine over 8 weeks in improving ADHD and ODD symptoms in children and adolescents with ADHD and comorbid ODD who had been non-responders to a previous parent support intervention.¹⁷⁹ Eligible participants were 6 to 15 years of age, who were diagnosed with ADHD and ODD according to DSM-IV criteria, and were required to have a score of at least 1.5 SD above the age norm for the ADHD subscale of the SNAP-IV, a CGI-S score of four or higher at screening and baseline, a SNAP-IV ODD subscale score of at least 15, and a normal intelligence score. All patients were provided open-label, parent support for 6 weeks. Patients who did not respond to the parent support phase (response was defined as an improvement in CGI-S score of 2 or more from baseline and at least a 30 percent decrease from baseline in ADHD sub-score of SNAP-IV) were randomized 3:1 to atomoxetine (n = 105) or placebo (n = 32) once daily for 8 weeks.

The atomoxetine dose was titrated from 0.5 mg/kg/day to a target dose of 1.2 mg/kg/day in 7 days. The primary efficacy measure was the ADHD subscale score of the SNAP-IV; the ODD subscale score of the SNAP-IV was a secondary outcome. Other outcome measures included health related quality of life as measured by means of the parent-rated child health and illness profile-child edition (CHIP-CE), and adverse events. Of the 156 patients who participated in the parent support phase, 139 were randomized and 137 were included in the efficacy analysis. Participants mean age was 9.7 (2.2) years in the atomoxetine arm and 10.0 (2.4) years in the placebo arm; 93 percent were males.

Mean baseline SNAP-IV ODD score was 17.5 for atomoxetine and 17.2 in the placebo arm. At the end of 8 week period the SNAP-IV ODD sub score significantly improved in the atomoxetine group compared to placebo [SNAP-IV ODD subscale score mean change: −2.7 (4.1) in the atomoxetine arm vs. −0.3 (2.6) in placebo arm, p=0.0001]. There was significant decrease in the ADHD subscale of the SNAP-IV in the atomoxetine arm compared to placebo. There was no significant differences between the mean changes of the CHIP-CE total score between atomoxetine (3.6) and placebo (1.2), p=0.071; however the atomoxetine group showed statistically significant differences compared to placebo for the subdomains of risk avoidance (p=0.013) and emotional comfort (p=0.007). The study was sponsored by industry.

Guanfacine (Intuniv)

A moderate risk of bias multicenter randomized, double blind, placebo-controlled trial conducted in the United States randomized children and adolescents with ADHD and oppositional symptoms 2:1 to receive either guanfacine extended release (n = 138) or placebo (n = 79) once daily for 8 weeks.¹⁷⁷ Eligible participants were between 6 and 12 years of age and had a DSM-IV-TR diagnosis of ADHD, a baseline score 24 or higher on the ADHD Rating Scale IV, and a baseline score 14 or higher (males) or 12 or higher (females) on the oppositional subscale of the CPRS-R:L. Following a 3-day to 5-week washout, participants underwent a 5-week dose optimization. During optimization, the dose of guanfacine extended release was increased in 1 mg/week increments to a maximum of 5 mg/day based on tolerance, the CGI-S score, and investigator judgment until the optimal dose was identified. Doses were maintained at the optimal level for 3 weeks. The primary outcome was change from baseline to endpoint in the oppositional subscale of the CPRS-R:L. Other outcomes included ADHD-RS-IV criteria, and adverse events. Participant mean age was 9.4 (1.7) years in the guanfacine extended release group and 9.3 (2.0) years in the placebo group. Mean score at baseline on the oppositional subscale of CPRS-R:L was 19.3 (4.74) in the guanfacine extended release group and 19.9 (4.29) in the placebo arm. Distribution of the optimal dose at the endpoint was: 1 mg (5.1%), 2 mg (27.2%), 3 mg (38.2%), and 4 mg (25%).

Guanfacine extended release significantly reduced oppositional symptoms as measured by the parent-rated CPRS-R:L oppositional subscale scores compared with placebo (least-square mean change from baseline: −10.9 for guanfacine extended release and −6.8 for placebo, p<0.001; effect size: 0.59) Least squares mean percentage reductions from baseline were significantly different between guanfacine extended release (56.3%) and placebo (33.4%) groups (effect size: 0.64, p<0.001). Significant reductions were observed in clinician-rated ADHD-RS-IV total scores in those treated with guanfacine extended release compared with placebo.