NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
Epstein R, Fonnesbeck C, Williamson E, et al. Psychosocial and Pharmacologic Interventions for Disruptive Behavior in Children and Adolescents [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2015 Oct. (Comparative Effectiveness Reviews, No. 154.)
Psychosocial and Pharmacologic Interventions for Disruptive Behavior in Children and Adolescents [Internet].
Show detailsState of the Literature
KQ1. Effectiveness of Psychosocial Interventions
Sixty-six studies examined the effectiveness of psychosocial interventions for children with disruptive behaviors. We categorized these studies broadly by age group as examining preschool (n = 23), school-age (n = 29), or teenage (n = 14) children, and according to whether the active treatment arm was an intervention that included only a child component (n = 2), only a parent component (n = 25), or was a multicomponent intervention (n = 39). Multicomponent interventions were defined as those that included two or more of a child component, parent component, or other component (e.g., teacher component, family together component). Studies within each of these intervention categories were heterogeneous, although several well-known programs were most common.
We included studies of interventions delivered in healthcare settings for children with a formal diagnosis of a disruptive behavior disorder or whose disruptive behaviors were assessed at or above a clinical cutoff on a well-validated measure of child disruptive behaviors. Thus, we excluded from our review studies of preventive or universal interventions, and interventions delivered in non-healthcare settings. These important interventions and populations may be appropriate for a separate review, but were beyond the scope of our review. We also excluded disruptive behaviors in the context of autism or other developmental disabilities. We included studies of children who had conditions such as attention deficit hyperactivity disorder (ADHD) as long as the primary focus of the study was on the treatment of the disruptive behavior. We also focused on parent reports of child disruptive behaviors because they were the most consistently reported outcome in the literature, because other outcomes of interest, especially functional outcomes such as school performance, were not consistently reported.
Preschool Children
Studies examining psychosocial interventions for preschool-age children had an active treatment arm that included only a parent component (n = 14) or were multicomponent interventions (n = 9). Seventeen of the 23 studies included in our review of psychosocial interventions for preschoolers with disruptive behaviors examined one of three interventions: Parent Child Interaction Therapy (PCIT) (n = 7), the Incredible Years Parent Training program (IY-PT) (n = 5), or the Positive Parenting Program (Triple P) (n = 5). The remaining six studies each examined a different intervention.
The seven studies examining PCIT for preschool disruptive behaviors evaluated several versions of PCIT (regular, abbreviated, culturally adapted) in comparison to treatment as usual, a waitlist control group, or another PCIT version. Although most studies measured child disruptive behaviors using the Eyberg Child Behavior Inventory (ECBI) Intensity and/or Problem subscales, most studies included other measures of child disruptive behaviors (e.g., Dyadic Parent-Child Interaction Coding System) and did not clearly identify one outcome measure as primary. All five of these studies reported significant reductions in parent-reported child disruptive behaviors from baseline to post-treatment in comparison to either treatment as usual or a waitlist control group, regardless of which version of PCIT was being evaluated. Consistent differences between versions of PCIT were not reported.
The five studies examining IY-PT for preschool disruptive behaviors evaluated several versions of IY-PT (IY-PT + ADVANCE, IY-PT, IY-PT psychologist led, IY-PT nurse led) in comparison to other versions of IY-PT and waitlist controls. All studies used one of the parent-reported ECBI scales or CBCL scales, and most of the studies included direct observation of child disruptive behaviors. On parent-reported measures of child disruptive behaviors, 5 studies reported improvement from baseline to followup (ranging from post-treatment to 2-year followup) for children in IY-PT. Children in the IY-PT arms consistently showed more improvement than children in waitlist control arms. Consistent differences between versions of IY-PT were not reported.
The five studies examining Triple P for preschool disruptive behaviors evaluated several different versions of Triple P against each other, a waitlist control group, and treatment as usual. Each of these studies reported significant reductions in disruptive behaviors in the Triple P treatment arm as compared to a waitlist control group on parent-reported child disruptive behaviors as measured by one of the ECBI subscales. Self-directed Triple P plus weekly phone conferences was found to be more effective than self-directed Triple P alone,135 and self-directed Triple P plus 14 hours of skills training and partner support was more effective than self-directed Triple P plus 10 hours of therapist-led skills training or self-directed Triple P alone.140
Although six other studies also examined interventions for preschoolers with disruptive behaviors, each examined a different individual intervention making it difficult to make general statements about these interventions.
Overall, most of the reviewed studies on psychosocial interventions for preschool children with disruptive behaviors focused on one of three specific interventions (PCIT, IY-PT, or Triple P). The literature for this age group is limited by difficulties defining the study population, study design limitations even among the RCTs, and lack of consensus about the most important outcomes.
School-Age Children
Seventeen of the 29 studies included in our review of psychosocial interventions for school-age children with disruptive behaviors had an active treatment arm that was a multicomponent intervention. Eleven studies included only a parent component and one study included only a child component. Four of the 15 studies of multicomponent interventions included at least two of the IY components (child training, parent training, and teacher training) in combination with one another, two were of a modular intervention, and two were of SNAP Under 12. The seven remaining studies were each of a different intervention.
Four studies examined more than one IY component in combination with each other. Because these studies test multiple IY component combinations against each other and waitlist control and measure multiple outcomes without designating a primary outcome, this group of studies is difficult to summarize succinctly. A conservative summary is that at least two IY components delivered together are associated with greater decreases in parent-reported child disruptive behavior than waitlist control.
Two studies (each including multiple papers) examined the effects of a community-based version (in comparison to a clinic-based version) or nurse-led version (in comparison to enhanced usual care) of a modular multicomponent intervention for children with ODD or CD. Both studies were therefore essentially testing the “portability” of this intervention. Although the nurse-led version was associated with improvement in goal achievement and overall health, it was not associated with significantly more improvement in parent-reported child disruptive behaviors than was enhanced usual care. Both the clinic- and community-based versions of the intervention were associated with significant reductions in parent-report child disruptive behaviors.
Two studies (one RCT, one non-RCT) compared the SNAP Under 12 intervention in comparison to a waitlist control group that engaged in recreational group activities. Children in SNAP Under 12 group in both studies showed greater reductions over treatment in parent-reported child disruptive behaviors as measured by the CBCL Aggression and CBCL Delinquency subscales. Only one study has been published for each of the remaining seven multicomponent interventions.
Of the 11 studies examining parent only interventions, three studies examined IY-PT, two studies examined PMTO, and six other studies each examined a different intervention with only a parent component. Two studies examined IY-PT in comparison to a waitlist control group. Each study reported significantly greater reductions on ECBI-I, ECBI-P, or both for children in IY-PT groups compared with the children in the control groups. Both of the studies examining PMTO reported that children receiving PMTO showed greater reductions in parent-reported child disruptive behaviors relative to treatment as usual. It is difficult to make general statements about the other six studies because they are each of a different intervention.
There was only one study including interventions with only a child component for school-age children. The study examined a social cognitive intervention program.132 As with the literature examining psychosocial interventions for preschool-age children, the literature on school-age children suggests that there is most support for multicomponent interventions that include a parent component. Overall limitations for the school-age literature are similar to that for preschoolers and are discussed in detail below.
Teenage Children
Thirteen of the 14 included studies examining psychosocial interventions for teenagers with disruptive behaviors had an active treatment arm that was a multicomponent intervention, specifically MST (n = 5) or BSFT (n = 3). The other three studies were each of a different intervention.
All five of the studies examining MST were RCTs. Two of these studies were conducted in the United States: one compared MST to treatment as usual;94 the other compared MST to individual therapy.111 Both of these studies demonstrated greater reductions in child disruptive behaviors for children receiving MST. The other three studies were conducted in Europe and compared MST to treatment as usual.120,124,136 One RCT compared youth randomized to receive MST with youth randomized to receive a treatment as usual intervention that was much more comprehensive than the type of treatment as usual most commonly included in studies conducted in the United States.120 This RCT reported that in comparison to youth randomized to the treatment as usual multicomponent intervention youth randomized to MST were less likely to have committed nonviolent offenses and experienced greater reductions in the CBCL Aggression and Delinquency subscales, but did not experience greater reductions in the CBCL Externalizing subscale from baseline to the end of followup.120 One RCT compared MST to treatment as usual reporting that youth randomized to receive MST experienced greater decreases in disruptive behaviors as measured by the CBCL Externalizing subscale, ODD and CD as measured by a DSM-IV symptoms checklist, and property offenses than did youth randomized to receive treatment as usual.136 The final of the RCTs conducted outside the United States compared MST to treatment as usual and reported no difference in disruptive behaviors as measured by the CBCL Externalizing subscale for youth randomized to receive MST as compared to youth randomized to receive treatment as usual.124
Three RCTs examined BSFT in comparison to group therapy106 or another family-based intervention.142,144 Although each of these studies examined the effectiveness of BSFT with a very specific subgroup (Hispanic teenagers)106 or very specific outcome (bullying),142,144 all three reported significant reductions in child behavior problems for the youth randomized to receive BSFT in comparison to group therapy106 or another multicomponent intervention.142,144
Although it is difficult to make general statements about each of the other four interventions included in this review because there is only one study of each, taken together the literature on psychosocial interventions for teenagers with disruptive behaviors suggests most support for multicomponent interventions such as MST or BSFT. Overall limitations are similar for teenage literature as for the two other age groups and are discussed in detail below.
Summary of Meta-Analysis
We conducted a Bayesian multivariate, mixed treatment (network) meta-analysis using data from RCTs addressing KQ1 that measured parent-reported child disruptive behavior using one of the most prevalent outcome measures (i.e., CBCL Externalizing subscale reported as a T-score, ECBI Intensity subscale, or ECBI Problem subscale) and included the necessary data at baseline and post-treatment for both intervention and control groups. In total, 28 studies were used to fit the model. The baseline was subtracted from the end-of-treatment mean and used as the response measure, along with the sum of their standard deviations. Our outcome variable was a standardized mean effect size. Our predictor variable was the broad category of intervention (child component only, parent component only, multicomponent) with the specific intervention type (PCIT, MST, etc.) defined as a random effect.
The effect sizes for the multicomponent intervention class and for interventions with only a parent component had the largest estimated value (Table 47), both with a median of −1.2 (95% credible interval: −1.6 to −0.9) standard deviations reduction in outcome score. The estimate for interventions with only a child component was −0.9 (95% Credible Interval: −1.6 to −0.4). Accordingly, multicomponent component interventions and interventions with only a parent component had the highest probability of being the best intervention (43% for both), followed by interventions with only a child component (14%).
Table 47
Network meta-analysis of intervention category as a predictor of treatment effect in parent-reported measures of child disruptive behavior among selected studies of psychosocial interventions.
Age effects were relatively more subtle, with an additive median effect of −0.4 standard deviations (95% credible interval: −0.6 to −0.3) for preschool relative to school-age children (baseline level), and of −0.1 standard deviations (95% credible interval: −0.5 to 0.2) for adolescents relative to school-age children. These trends were evident across each of the outcome measures included in the analysis.
The marginal posterior probabilities of remaining above the cut point were higher for the treatment as usual/control group relative to each intervention group, with multicomponent interventions showing the lowest proportion of children still above the clinical cutoff post-treatment.
Though we considered age-by-treatment interactions, there was not enough balance among the age and treatment combinations to include them in the final model.
Overall Summary of KQ1
Our qualitative and quantitative syntheses generally suggest that available evidence provides the most support for interventions for children with disruptive behaviors that are multicomponent interventions or interventions that include only a parent component. All multicomponent interventions included in this study included a parent component. Our assessment of the overall strength of evidence and limitations of this evidence base are discussed in detail below and in an Evidence Profile (Appendix J). Overall, the evidence base is limited by difficulty defining the study population, study design limitations even among RCTs, and lack of consensus about the most important outcomes.
KQ2. Effectiveness of Pharmacologic Interventions
Despite a fairly robust literature on psychopharmacologic drugs as a whole, few studies have focused specifically on their use in children whose primary indication is a disruptive behavior disorder. Among those studied are four types of drug treatment: antipsychotics, antiepileptics, stimulants typically used with ADHD, and nonstimulants typically used with ADHD. Thirteen studies were identified across all of these drug classes, with the most commonly studied drug being risperidone. All studies were conducted in primarily male patient populations, with ages ranging from 6 to 18 years.
Four antipsychotics were studied: risperidone, quetiapine, aripiprazole, and ziprasidone. All of the antipsychotics are second-generation atypical antipsychotics. Prior systematic reviews have studied these drugs and others for a more general set of indications. We describe the findings of those reviews in below in order to place our more limited review in context. Four antipsychotics were studied: risperidone, quetiapine, aripiprazole, and ziprasidone. All of the antipsychotics are second generation, atypical antipsychotics. There is a large literature base as well as prior systematic reviews that have studied these drugs and others for a more general set of indications and are available.
Among antipsychotics, risperidone was assessed in three studies and the others were each in only one study. Among studies of risperidone, two studied the effectiveness of risperidone as the initial treatment, and one focused on maintenance, comparing continued use of risperidone to discontinuation and replacement with placebo. One study186 reported a positive effect of risperidone over placebo. The second study181 compared risperidone as augmentation to stimulant medication for patients with ADHD and aggressive behavior after treatment with a stimulant and found no benefit of risperidone over placebo. The final study183 was a maintenance study, comparing risperidone to placebo after treatment with risperidone. In this study, the placebo group worsened more than the risperidone group over 6 months.
One open label study224 compared aripiprazole to ziprasidone and found the two medications to be equally effective in decreasing clinically significant aggressive behavior over two months.
Finally, one study compared the use of quetiapine to placebo and the results were mixed. Although clinicians rated greater improvement in symptoms in the quetiapine group, there was no difference on parent rated measures, and no difference in CPRS (a rating of general problem behaviors in children). Quality of life was reported to be significantly higher in the treatment group, however.
Overall, these studies were limited by short duration (all but one183 were 2 months or less) or high attrition. While head-to-head studies such as that comparing aripiprazole to ziprasidone are useful to compare medications, large, randomized, controlled studies that measure effect size and show consistent benefit of this class of medications over placebo are also needed.
Among antiepileptic drugs, only valproic acid was studied specifically for disruptive behavior disorders. The one placebo-controlled trial178 reported no benefit for valproic acid, while a small (n = 20) crossover study of slightly older (up to age 18) children reported a benefit for drug. Finally, one study provided valproate to all participants, but compared high and low doses, with greater effects reported for higher doses and in “high distress” conduct disorder relative to “low distress” conduct disorder. All three studies were small, short-term and funded by the manufacturer of the treatment drug.
Medications commonly used for ADHD, both stimulants and nonstimulants, have also been studied for their potentially specifically to manage disruptive behaviors among children with comorbid ADHD and ODD/CD. Among the nonstimulants, Atomoxetine, a centrally acting, norepinephrine reuptake inhibitor, has been approved for treatment of ADHD in children and adolescents and has been used off label for treatment of DBD and ODD symptoms among populations with comorbid ADHD.
Two RCTs176,179 reported treatment with atomoxetine (up to 1.2 mg/kg/day) for nine weeks improved ODD symptom scores compared to placebo, among children and adolescents with ADHD and comorbid ODD. One RCT (reported in 2 publications)176,190 reported significant improvement in quality of life compared to placebo, over the 9-week period. The other RCT179 found no significant differences in overall quality of life, but improvement in certain subdomains, including risk avoidance and emotional comfort.
Guanfacine extended release is a selective central alpha2A-adrenergic receptor agonist and is FDA approved for treatment of ADHD in children 6 to 17 years. We identified one low risk of bias RCT of guanfacine extended release (1 to 4 mg/day) that reported significantly reduced oppositional symptoms as measured by the CPRS-R:L oppositional subscale scores compared with placebo.
Although there were a limited number of studies, these three RCTs reported short-term effectiveness in reducing ODD/CD symptoms among children and adolescents with comorbid ADHD and ODD.
One high risk of bias RCT of mixed amphetamine salts extended release demonstrated that higher doses (30 mg/day) were associated with decreased ODD symptoms compared to placebo over a 4-week period among school-aged population with ODD, 79 percent of who also met criteria for ADHD. This study also reported significant improvement in several quality of life measures for children with ODD. One high risk of bias RCT of methylphenidate (up to 60 mg/day in 2 divided doses) among school-aged population with CD symptoms, 69 percent of who also met ADHD criteria, found both teacher and parent ratings of CD problems improved compared to those in the placebo group. Duration of these two studies was short, ranging from 4-5 weeks, which is too short to determine whether there is a long-term treatment effect of stimulants on ODD symptoms. Severity of ODD symptoms at baseline may be important mediator in treatment response, but more data are needed to examine this question.
Overall, studies are lacking that compare children/adolescents with ADHD alone to those with ADHD with ODD and ODD alone in order to evaluate the specific effects of treatment on oppositional symptoms. Comorbid ODD is commonly present in children and adolescents with ADHD, and thus finding populations with ADHD but without ODD symptoms or ODD without ADHD may be challenging. Most importantly, it is unclear whether treatment-related changes in ODD symptoms are independent of changes in ADHD symptoms in this population. Treatment period of 8 to 9 weeks may be too short to determine whether there is a long-term atomoxetine treatment effect on ODD symptoms or quality of life outcomes.
Although combination therapy with antipsychotics and stimulants can be effective for patients with ADHD and comorbid DBD, we found a lack of studies that evaluated combination pharmacologic treatment compared to monotherapy or compared the efficacy of combined behavioral and pharmacologic interventions compared to pharmacologic or behavioral interventions alone. To date, treatment research is almost exclusively supported by the pharmaceutical industry. Given the prevalence of DBDs and the need for high quality data to inform clinical practice, more long-term studies are needed.
KQ3. Effectiveness of Psychosocial Versus Pharmacologic Interventions
No head-to-head studies were found to answer this question.
KQ4. Effectiveness of Combined Psychosocial and Pharmacologic Interventions
No head-to-head studies were found to answer this question.
KQ5. Harms of Psychosocial or Pharmacologic Interventions
No harms of psychosocial interventions were reported. Importantly, a lack of reported harms is not an indication that no harms exist. The psychosocial literature also uniformly failed to note that harms were sought.
The medical treatment studies in this report were generally small and short term, with typically no followup post treatment. Thus, harms reported in those studies were generally mild or moderate and fairly immediate in nature. Nonetheless, there was significant loss to follow up in several studies, some of which was clearly due to experiencing adverse events and the studies were very short term and not powered to identify harms that might be rare. All of the pharmacologic studies included in the empirical literature here were designed and powered for benefit and thus would only be likely to identify common and minor events. Therefore, we sought harms data from other sources that might include more extensive and longer-term data, including other systematic reviews. It is important to note that other studies, including large scale, database analyses have identified harms of antipsychotics in particular to include significantly increased risk of metabolic effects.
Harms of the antiepileptic drug, valproate, were available from three, short-term RCTs. They include short-term changes in sleep pattern, mood changes, gastrointestinal upset and appetite changes. The proportion of patients experiencing these were high some cases (e.g. 50% with insomnia in the treated group versus 15% in placebo), but the numbers of participants included were so low as to likely be unstable. Additional data are available from FDA package inserts and provide more support for these adverse events. Specifically in pediatric clinical trials, consisting of 76 patients aged 10 to 17 years taking divalproex extended release for mania and 231 patients aged 12 to 17 years taking divalproex extended release for migraine, common adverse events (reported more than 5% and twice the rate of placebo) included: nausea, upper abdominal pain, somnolence, increased ammonia, gastritis and rash.211
Harms of antipsychotics have been studied extensively, both in the context of effectiveness research and independently, and are reviewed in other systematic reviews. The systematic review data mirror that available in our included studies, namely identifying significant increases in somnolence, fatigue and weight gain.183
We identified three good quality systematic reviews addressing harms of atypical antipsychotics in children and adolescents.49,52,198 Harms found to be significantly associated with treatment included weight gain and changes in metabolic parameters.49 Mean weight gain in the risperidone group was 2.37 kg more than in the placebo arm over 6 to 10 weeks in a meta-analysis of two trials (mean difference: 2.37, 95% CI: 0.26 to 4.49).
Another Agency for Healthcare Research and Quality (AHRQ) review included studies of atypical antipsychotics used for any indication in individuals aged 24 years and younger.52 Agents included in studies in the review were haloperidol, risperidone, aripiprazole, olanzapine, pimozide, quetiapine, clozapine, and ziprasidone, and median study duration was 8 weeks. The review evaluated harms by drug class and noted fewer extrapyramidal symptoms associated with olanzapine and risperidone compared with haloperidol (low strength of the evidence), and no significant differences between first and second-generation antipsychotics in prolactin-related adverse events (low strength of the evidence). Risperidone was associated with less dyslipidemia and less weight gain that olanzapine (moderate strength of the evidence). Risperidone was also associated with more prolactin-related harms than olanzapine (moderate strength of the evidence) and with more weight gain than aripiprazole (low strength of the evidence).
Finally, one review and meta-analysis evaluated metabolic and neurologic adverse events associated with second-generation antipsychotic use in children with any mental health disorder and included 35 RCTs (4 reported in the current review).198 In a meta-analysis of 10 RCTs of risperidone of less than 12 weeks duration, weight gain (mean difference: 1.72 kg, 95% CI: 1.17 to 2.26, p<0.00001), prolactin levels (mean difference: 20.70 ng/mL, 95% CI: 16.78 to 24.62, p<0.00001), and change in prolactin from baseline to end of treatment (mean difference: 44.57 ng/mL, 95% CI: 32.24 to 56.90, p<0.00001) were higher in risperidone groups compared with placebo. The odds of clinically significant weight gain were higher in the risperidone arm compared with placebo (OR=2.90, p=NS) as were the odds of extrapyramidal symptoms (OR=3.35, p<0.0001). The review reported no clinically significant changes in laboratory values or blood pressure in seven studies. Blood pressure was elevated in the risperidone group in one study. In studies comparing risperidone at different doses or with other agents (pimozide, clonidine, haloperidol), children in the risperidone arms had weight gain and extrapyramidal symptoms that were typically not significantly different from the comparison group, though higher doses of risperidone were associated with greater weight gain and movement symptoms.
In a meta-analysis of three studies of quetiapine versus placebo (including the Connor 2008180 RCT included in the current review), weight gain, but not prolactin levels, was significantly higher in the quetiapine group (mean difference: 1.41 kg, 95% CI: 1.01 to 1.81). Triglyceride levels, blood pressure, and heart rate were significantly elevated in the quetiapine group compared with placebo in one RCT. The review also included nine RCTs assessing aripiprazole, five of which were combined in meta-analyses. Mean weight gain (mean difference: 0.85 kg, 95% CI: 0.57 to 1.13, p<0.00001) and BMI increase (mean difference: 0.27 kg/m2 95% CI: 0.11 to 0.42, p=0.0007) were higher in aripiprazole groups compared with placebo, and the odds of weight gain were significantly higher in the treatment group (OR=3.66, p=0.0003). Lipids and ECG values did not differ significantly between groups.
Significant risk of metabolic effects has also been demonstrated to be elevated in large database analyses.225 Indeed, given the methodologic challenges to reviewing harms in RCTs that are noted above, observational studies such as these are likely to provide more precise estimates of harms. They unfortunately are not necessarily limited to the population of interest in our review, and provide little detailed clinical information.
Harms of stimulants, including methylphenidate and amphetamine salts, which are typically used to treat ADHD and commonly used for DBD, include delay of sleep and anorexia, particularly at higher dosage in the included study. The FDA package insert includes a warning that methylphenidate has been associated with sudden cardiac death in children with existing cardiac abnormalities.
Nonstimulants, including atomoxetine and guanfacine, were associated with increased rates of headache, somnolence, and anorexia.
None of these studies described here, however, explicitly weigh the benefits achieved against these harms, and clinicians and families need to do so including both effectiveness and harms evidence.
KQ6. Modifiers of Effectiveness of Interventions
This question was divided into sub-questions about variations in intervention effectiveness due to a) patient characteristics, b) characteristics of the disorder, c) patient treatment history, and d) treatment characteristics. Although studies examining each of these questions were identified, it is unclear if any of the identified studies were adequately powered to answer these questions.
Regarding variations in the effectiveness of psychosocial interventions due to patient characteristics, it is important to note that most studies included relatively homogeneous age groups (e.g., preschool, school, or adolescent children). That studies were restricted to specific age groups implies widespread, tacit acceptance of the idea that intervention effectiveness varies by child age. At the same time, this aspect of study design limits the ability of included studies to examine this issue. The most commonly examined patient characteristics include child gender, characteristics of the child's mother, and characteristics of the child's family. In general, results were inconsistent and additional examination of these issues is warranted. None of the studies examining pharmacologic interventions addressed the potential for variations in treatment effectiveness based on patient characteristics.
The most commonly examined characteristic of disruptive behavior disorders that was examined for its potential to moderate the effectiveness of psychosocial interventions is the severity of a child's disruptive behaviors at baseline. Results were inconsistent. Difficult temperament and psychopathy were associated with treatment effectiveness in studies with pre-kindergarten age children and studies with teenage children, respectively. More examination of these characteristics is needed.
The severity of a child's disruptive behaviors at baseline and the presence ODD or CD comorbid with ADHD were the characteristics of the disorder that studies of pharmacologic interventions were most likely to examine for their association with differential treatment effectiveness. In general, more disruptive behavior at baseline was associated with greater treatment effectiveness. It is unclear, however, if changes in non-ADHD disruptive behaviors are independent of changes in ADHD symptoms because of the high prevalence of comorbidity in the study populations.
No studies evaluated variations in the effectiveness of psychosocial interventions due to patient treatment history. One study of atomoxetine indicated that prior treatment with a stimulant was associated with a larger treatment response to atomoxetine.
Studies of psychosocial interventions evaluated variation due to treatment characteristics based on dose – defined by some measure of treatment attendance – and, for interventions including a parent component either alone or in combination with other components, based on whether changing parenting practices mediated intervention effectiveness. The studies defining dose either as session attendance or as homework completion consistently reported greater intervention effects for children whose parents participated more. Similarly, studies examining whether changes in parenting practices were associated with treatment effectiveness consistently provided some support that they were. This is consistent with results from prior reviews.226
In pharmacologic studies, the role of baseline severity was inconsistent, with no mitigating effect of severity for nonstimulants, but greater effect associated with greater baseline severity in one RCT of the stimulant mixed amphetamine salts ER.182 It is not clear if treatment-related changes in ODD symptoms are independent of changes in ADHD symptoms in this population. One study of atomoxetine176 used a path analysis to evaluate if the treatment effect on ODD symptoms were influenced through the treatment effect on ADHD and/or CD symptoms; they found a nonadditive effect, implying a negative direct effect of atomoxetine on ODD symptoms. In a post hoc analysis of another atomoxetine RCT, authors found that the percent reduction from baseline to endpoint in oppositional symptoms (CPRS-R:L ODD subscale) and ADHD symptoms were highly correlated (r=0.74).
Findings in Relationship to What Is Already Known
We searched for systematic reviews published between 2005 and 2014. We evaluated each for relevance to our Key Questions using the review PICOTS (Appendix B).We identified 22 reviews assessing the effectiveness of psychosocial interventions and two reviews assessing the effectiveness of pharmacologic interventions.
The reviews of psychosocial interventions included two types: reviews of literature regarding specific interventions such as MST and reviews of more general interventions. These reviews did not address potential harms of psychosocial interventions. The two reviews of pharmacologic interventions addressed the effectiveness of atypical antipsychotic medications, though they were not specific to populations of children treated for disruptive behaviors. We describe information about harms from these two reviews (and one additional review that reported harms only) in KQ5 above.
Existing Reviews of Psychosocial Interventions
Of the 22 identified systematic reviews or meta-analyses of psychosocial interventions, we identified one review specific to the MST literature, one review specific to the CBT literature, two reviews specific to the Triple P literature, one review of Triple P and PCIT, 11 more general reviews, and one review of existing reviews. It is important to note that these reviews may include studies not included in the current review due to different study inclusion and exclusion criteria.
Reviews of MST Literature
A Cochrane review included eight RCTs of MST for behavioral and emotional problems in children between the ages of 10 and 17 years.227 Few studies addressed outcomes related to externalizing behaviors, but an analysis of three studies reporting CBCL Externalizing scales showed pooled results were not significant (standardized mean difference: −0.18, 95% CI: −0.46 to 0.09). This review concluded that there is inconclusive evidence of the effectiveness of MST as compared to other interventions.
Reviews of CBT Literature
One meta-analysis included six studies of CBT for violent behavior and reported limited effects of CBT on child behavior (effect size: −0.094), with decreasing effects reported over time as outcome data accumulated.228 In a cumulative analysis, effect sizes decreased from 0 to −0.95. This review concluded that this is a medium effect and called for more research into the effectiveness of CBT for violent behavior.
Reviews of Triple P Literature
There were two reviews of the Triple P literature.229,230 Both of these reviews were meta-analyses and together reported on more than 100 randomized, nonrandomized, and uncontrolled studies.229,230 These reviews both reported small to medium effect sizes for parenting outcomes (0.38 to 0.57), child behavior outcomes (0.35 to 0.52), and parental wellbeing/satisfaction (0.17 to 0.55) over both the short- and long-term.
These reviews also examined potential moderators. One of these reviews examined if factors related to the specific design of the version of the Triple P program that was implemented (group vs. individual) and reported no consistent effect modification by design.229 This review also reported that treatment effects were consistently lower when measured by father reports than those of mothers, and that younger child age was associated with greater treatment effectiveness. Neither level of initial behavioral severity nor child gender were significant predictors of outcomes. Program completion was also not associated with the effect size.229
The other review reported that Triple P level and Triple P as a treatment (vs. Triple P as a preventative intervention) were associated with higher effect sizes.230 Online Triple P had the largest effect sizes for child outcomes; the online and group versions of Triple P had the largest effect sizes for parental relationship outcomes. Study power also moderated treatment effects such that higher effect sizes were found for studies with less than 35 participants in the smallest group compared to studies with greater than 35 participants in the smallest group. Significant effect sizes were found for studies with larger sample sizes. This review also reported that the initial severity of child behaviors moderated effects on parental relationship outcomes. This review also noted that Triple P studies that did not include involvement of a developer of the program (n = 31) still produced significant intervention effects on child outcomes.
Reviews of Triple P and PCIT
One meta-analysis examined the effects of PCIT (n = 13 studies) and Triple P (n = 11 studies) on parent-reported child problem behaviors.231 Children were in the clinical or borderline range for disruptive behaviors in most (but not all) of the studies, and effect sizes in studies reporting between-group comparisons ranged from −1.59 to 5.67 across parent, teacher, father, and observation measures and control groups for PCIT. All forms of PCIT except the abbreviated version had a significant short-term effect on parent-reported child behaviors.
Effect sizes ranged from −0.96 to −0.02 for Triple P across informants and all formats (group, self-directed, etc.). The effect size for PCIT (−1.45) was significantly higher than those for self-directed (−0.51), group (−0.67), and individual Triple P (−0.69) but not for media or enhanced Triple P. Effects sizes for observed outcomes were not significantly different from PCIT and any form of Triple P. Negative parenting behaviors were also improved with both PCIT and Triple P. The review notes that limited evidence addresses effects over the longer term.
More General Reviews
There were 11 general reviews, not restricted to literature about specific interventions. These reviews included RCTs, controlled trials, and quasi-experimental studies published in any country, including studies dating to the 1980s. As in the current review, participants in the studies included in prior systematic reviews were mostly male and typically Caucasian. Most of the included studies were small, with short-term (≤ 6 months) followup. Studies were generally of moderate methodological quality, with reporting of family characteristics, allocation concealment, and randomization methods generally noted as limited. Reviews described variations in inclusion criteria (e.g., requirement of DSM diagnosis of a DBD, only parent-reported problem behaviors) and recruitment methods. A brief summary of each of these reviews is included below.
One meta-analysis included 28 RCTs for a total of 2239 children with disruptive behaviors between the ages of 2 and 12 years.59 Fourteen studies assessed variations of the IY program, nine studies evaluated the Triple P, two studies evaluated PCIT, and three studies assessed other approaches such as “Project TEAM.” The investigators rated six of the studies as low or moderate risk of bias. Overall, the studies varied in terms of how they defined their target population with studies using clinical cut-off levels on established measures, DSM diagnoses, or general parent-reported behaviors to establish inclusion. Reporting of parent and family characteristics also varied across studies. Child disruptive behaviors measured on the ECBI Intensity and Problem scales were significantly (p<0.001) reduced in active treatment arms compared with control (weighted mean differences of −20.90, 95% CI: −26.26 to −15.53 and −6.03 95% CI: −7.70 to −4.36, respectively). CBCL Externalizing scores and Strengths and Difficulties Questionnaire (SDQ) Conduct scales were also significantly improved in active treatment arms compared with control groups.
Another meta-analyses of 57 RCTs of parenting programs for child disruptive behaviors reported similar results.55 In combined analyses of ECBI-Intensity and CBCL scales reported in 24 of these studies, children in the parent management training intervention arms had improved outcomes compared with children in comparison arms (standardized mean difference: −0.67, 95% CI: −0.91 to −0.42). Investigators' analysis of other outcomes reported in 36 studies aligned with these meta-analysis findings: for 100 of 170 child behavior outcomes, outcomes were significantly improved in children in treatment groups compared with those in control groups. Meta-analysis of independent observations in seven studies also demonstrated significantly improved outcomes for children in active treatment groups versus control groups (standardized mean difference: −0.44, 95% CI: −0.66 to −0.23).
A United Kingdom National Health Service review of 37 RCTs of parenting interventions for children with conduct disorder also reported consistent evidence for the short-term effectiveness of parent training programs compared with control groups.232 Six included studies were assessed as good or adequate quality; the other 31 studies were rated as poor or very poor quality. Pooled estimates demonstrated significant improvement in treatment groups compared with control on the ECBI, CBCL, and in observer coding of parent-child interactions, while differences between the parent management approaches studies were not consistent.
Another review of seven studies of primarily parent management approaches addressed intervention for children with ODD and reported the greatest effects on child behavior when interventions targeted parents (standardized mean difference: 1.06; 95% CI: 0.70 to 1.41), with smaller effects if only children were targeted (standardized mean difference: 0.93; 95% CI: 0.52 to 1.34).233
A meta-analysis including 79 studies reporting on children with externalizing behaviors noted a mean weighted effect size of 0.30 (95% CI: 0.21 to 0.39) for end of treatment child behaviors in studies with comparison groups, 0.68 (95% CI: 0.59 to 0.77) for within group comparison studies, and 0.54 (95% CI: 0.43 to 0.65) in single subject studies.234 Effect sizes at followup were 0.40 for between-group designs, 0.79 for within groups, and 1.74 for single subject designs. Modifiers across each study type varied, but child age, method of treatment delivery, use of randomization, use of reliability assessment, and number of treatment sessions were significant modifiers of effects in between-group studies, with studies that included children age 9 to 11 years (n = 2) had larger effects. Those using individual consultation and controlled learning and those using non-random assignment also had larger effects as did those not reporting a reliability assessment. Finally, studies using between one and five treatment sessions had greater effect sizes than those using more sessions.
Another meta-analysis of 63 studies including children with DBD reported overall effect sizes of 0.42 for child behavior outcomes, 0.47 for parent behaviors, and 0.53 for parental perceptions in the short term and smaller effect sizes in the longer term.235 Children from families with lower socioeconomic status had less improvement of behaviors than did the children from families with higher socioeconomic status (p<0.01), as did those in studies with groups with more single parents compared to those with fewer single parents (p<0.01). Children with clinically significant baseline levels of disruptive behavior had greater change than did children without such clinically significant behaviors (p<0.05). Socioeconomic status also significantly moderated parent outcomes, with lower socioeconomic status associated with poorer outcomes. In contrast to our findings, treatment modalities involving only the parent were associated with greater positive change than those delivering interventions to the child separately or using a multisystem approach (p<0.05). Change in parent perceptions (confidence, stress) was also greater with parent-only interventions compared with those involving parents and children (p<0.05).
One meta-analysis included 33 studies of psychosocial interventions (including but not limited to behavior therapy, family therapy, CBT, psychodynamic therapy) with untreated comparison groups.236 These studies included many of the IY and PCIT studies also included in the other meta-analyses, but this review did not separately report results for any specific intervention group. Effect sizes in all 33 studies indicated improvement after treatment in active treatment vs. control arms with an overall mean weighted effect size of 0.62 (95% CI: 0.49 to 0.76). Smaller sample sizes were associated with larger effect sizes as compared to studies with larger sample sizes.
Another meta-analysis of broadly defined psychosocial interventions (including behavioral approaches and non-behavioral approaches such as family systems interventions and nondirective counseling, and named interventions such as variations of IY and PCIT) included 36 RCTs (n = 3042 children).237 The overall effect size (effect sizes for aggression, oppositionality, impulsivity, and general externalizing behaviors combined) for psychosocial treatments on disruptive behaviors was 0.82 (SE=0.10, 95% CI: 0.63 to 1.01). Significant moderators of effect included symptom type, with externalizing symptoms showing the largest response. Treatment as usual comparators also yielded larger effects than no treatment control groups, and no treatment controls yielded greater effects than “education, support, and attention” controls. Behavioral treatments demonstrated larger effects on behavioral symptoms than did non-behavioral interventions.
A review including 28 studies including 16 psychosocial interventions for children with disruptive behavior disorders for which there was an evidence base. The authors summarized their results by classifying interventions according to Chambless criteria,238 as “well established” (e.g., MPTO), “probably efficacious” (e.g., Group Assertiveness Training (peer and counselor led), Anger Control training, Helping the Noncompliant Child, IY (child, parent, and multiple component), Multidimensional Treatment Foster Care, PCIT, Problem Solving Skills Training (multiple versions), Rational-Emotive Mental Health Program, and Triple P standard and enhanced programs), or “possibly efficacious” (e.g., IY with teacher training components, Triple P standard group treatment, First Step to Success, Reaching Educators, Children, and Parents, Self-Administered Treatment Plus Signal Seat, and Group Anger Control Training). The investigators recommend parent training as the first line treatment for young children and that direct child training or multicomponent approaches be used with older children.
A meta-analysis including 71 studies of interventions categorized as either parent management training or CBT reported positive outcomes associated with parent management training interventions.239 The mean effect size for both interventions combined (also combining parent and teacher-reported measures and observation outcomes) was 0.40 (95% CI: 0.34 to 0.47). The mean effect size for parent management training alone was 0.47 (95% CI: 0.34 to 0.61) and 0.35 (95% CI: 0.25 to 0.47) for CBT alone. In comparisons of effect sizes in children between ages 6 and 12 (ages were too widely varying to allow other comparisons), the effect size for parent management training (0.45, 95% CI: 0.28 to 0.60) was significantly greater than that of CBT (0.23, 95% CI: 0.11 to 0.32) in analyses not controlling for intervention setting (e.g., clinic, school). The difference was not significant in analyses controlling for setting. In moderator analyses, child age was not significantly associated with outcomes of parent management, but older child age was associated with better outcomes in studies of CBT.
A Campbell Collaboration review and meta-analysis included 55 RCTs including children age 5 or younger and focused on prevention of child behaviors such as delinquency, crime, and antisocial behavior.240 Most studies (n = 47) evaluated programs included in the current review such as IY variations, PCIT, and Triple-P. Eight trials assessed home visits by clinicians. Most studies were in the United States (n = 38), and most (n = 37) included fewer than 100 children. The overall weighted mean effect size across all 55 studies was 0.35, a small to medium effect for reducing child behavior problems. Differences between parent training and home visit programs were not significant. In regression analyses, older studies, smaller studies (n<100), and U.S. studies were more likely to have larger effect sizes. Meta-analyses also suggested the presence of publication bias.
Reviews That Include Existing Reviews
Findings in a review of RCTs, quasi-experimental studies, and systematic reviews and meta-analyses conducted for the U.S. Department of Health and Human Services, Substance Abuse and Mental Health Services Administration's (SAMHSA) Assessing the Evidence Base series of literature reviews also align with findings in prior syntheses.241 The SAMHSA review reports a high level of evidence (defined as confidence in the reported outcomes based on three or more well-conducted RCTs or two RCTs and well-conducted quasi-experimental studies) for both IY and PCIT, noting well-designed RCTs of adequate power, manualized approaches, reliable outcome measurement, and replication in multiple studies. IY and PCIT were also associated with improved externalizing behavior outcomes across age ranges and populations when compared with waitlist control groups, and the review concludes that abbreviated or adapted versions of IY and PCIT are also promising.
Summary of Evidence From Existing Reviews on Moderators and Mediators of Effectiveness
A number of existing reviews examined questions related to moderation and/or mediation of intervention effectiveness including a wide range of demographic and clinical variables. Most of the examined demographic and family process variables were not consistently identified as moderators with the strongest evidence appearing to include severity of baseline child disruptive behaviors, child age, and socioeconomic status.
Regarding the severity of baseline child disruptive behaviors, three existing reviews presented evidence suggesting that severity moderates intervention effectiveness, 230,235,242 and one review did not.229 One quasi-systematic review identified six studies assessing baseline child behavior as a moderator. In four of these six studies, higher baseline levels of problem behavior were associated with better outcomes but the other two studies did not.243 Two previous reviews cited evidence that child age moderated intervention effectiveness, 229,234 and two studies reported that family socioeconomic status also moderates intervention effectiveness.242,244
Finally, one quasi-systematic review specifically examined if parenting was a mediator of the effectiveness of behavioral parent training for child disruptive behaviors and provided some, but not overwhelming, support for this hypothesis.226
Existing Reviews of Pharmacological Interventions
We identified two reviews of the effectiveness of pharmacologic interventions for children, though not all of the included studies were specific to populations of children treated for disruptive behaviors. We describe information about harms from these two reviews (and one additional review that reported harms only) in KQ5 above.
One Cochrane review of atypical antipsychotics for disruptive behavior disorders included eight RCTs (7 of risperidone and 1 of quetiapine) and reported limited evidence of effectiveness. In one analysis, scores on the Aberrant Behavior Checklist were 6.49 units lower, which may be clinically significant, and the investigators considered the difference of 8.61 points on the Nisonger Child Behavior scale as likely clinically significant.49
In the other included review of pharmacologic interventions, an AHRQ-funded review of antipsychotic use in children and young adults, strength of the evidence was insufficient for comparisons of first versus second-generation antipsychotics, first versus first generation, and first generation versus placebo.52 In eight studies of antipsychotics for treatment of disruptive behavior disorders (including 682 children treated for between 4 weeks to 6 months), strength of the evidence was moderate for positive effects of antipsychotics on behavior symptoms and low for positive effects on aggression and anxiety.
Applicability
KQ1. Psychosocial Interventions
Applicability for this literature is largely dependent on the target population and feasibility of the interventions in real-world clinical settings. Our target population was primarily defined by child age and type of disruptive behavior problem. Included psychosocial interventions excluded preventive interventions, were typically multi-faceted and heterogeneous within broad intervention categories, and can be resource intensive relative to time, money, and personnel in the clinical setting.
Approximately half of the studies of psychosocial interventions for child disruptive behaviors were of school-age children, about 30 percent were with preschool-age children, and approximately 20 percent were with teenagers. We defined a study as focusing on school-age children if it had a sample with a mean age between 5 and 12 years. We established 5 years of age as the lower bound because this is the age at which children typically begin attending kindergarten in the United States. We established 12 years of age as the upper bound because 13 years is regarded as the beginning of adolescence in casual parlance. For precisely these reasons, the age group classification we used is somewhat arbitrary, specific to the United States context, and has face validity in the United States. At the same time, many studies of child samples with a mean age between 5 and 12 years also included children with age less than 5 or greater than 12 years.
In addition to the age definition, our definition of the target population included children with disruptive behaviors receiving treatment in healthcare settings. We did not restrict our study population to children meeting formal diagnostic criteria for a disruptive behavior disorder. Rather, we allowed children without a diagnosed DBD but with disruptive behaviors above a measure-specific threshold on a well-validated measure of disruptive behavior to be included. This may limit applicability of our findings because in real-world clinical settings third-party payers may only reimburse for services regarded as medically necessary. We excluded studies of preventive interventions for an at-risk population because our review was focused on studies of individuals who met a clinical threshold for a disruptive behavior disorder.
A potential issue for applicability of these findings is whether patients are able to access and pay for them if insurance does not cover them. However, an evaluation of costs was beyond the scope of this report as it was set up. Applicability of our findings is also limited by restricted access to some of the interventions most commonly examined in the studies included in this review in real-world clinical settings. Many of the included studies were conducted in the outpatient setting and carried out at academic medical centers in the United States. To give just one example, although there was relatively strong evidence in favor of the effectiveness of MST for disruptive behaviors in teenagers, MST is often not available in real-world clinical settings. This is consistent with a growing literature on the challenges of transporting evidence-based multicomponent interventions into real-world clinical settings with fidelity.
Many included studies were also conducted by the intervention developer or by other individuals with a vested interest in the intervention. Although this aspect of study design may be required to ensure treatment fidelity or at least make it more likely that interventions are delivered with fidelity to the model, it may also create a need for independent validation of study results.
KQ2. Pharmacologic Interventions
The populations studied the studies of pharmacologic interventions for disruptive behavior disorders were almost exclusively male and between the ages of 6 and 18 years. All of the studies were very small, and results may not be broadly generalizable. None of the interventions have a specific indication for disruptive behaviors, although they are used for these conditions in the United States. Interventions included antipsychotic drugs, an antiepileptic drug, and ADHD drugs (both stimulants and nonstimulants). Of particular importance, all but one study on pharmacologic interventions were funding wholly or partially by a pharmaceutical company, or were conducted by individuals who are highly supported by those companies. It is difficult to assess the degree to which these drugs are or are not widely available. The studies also did not address the common concern of polypharmacy and thus there may be limited ability to assess applicability as well as safety concerns in highly complex cases. Polypharmacy with two or more antipsychotic drugs is a commonly used indicator of poor quality care although it clearly occurs. A better understanding of the prevalence, circumstances, and implications of polypharmacy is needed.
In reality, many if not most children and adolescents seeking treatment for disruptive behaviors may have multiple co-diagnoses and other complex challenges. The applicability of this set of studies, in which we limited the population to a specific focus on disruptive behavior treatment, may not capture the overall effect of pharmacologic intervention on these children's lives overall, nor are they likely to be applicable to highly complex cases. The use of pharmacologic interventions for outcomes in cases, for example, of ADHD, autism or other conditions like bipolar is addressed in other reviews.
Strength of Evidence
We assessed strength of evidence for the effectiveness of interventions using the qualitative and quantitative approaches described in the Methods section. Overall, the evidence to answer Key Questions about interventions for children with disruptive behavior disorders was insufficient to moderate. We summarize the strength of the evidence and provide the assessment of the risk of bias, consistency of findings across trials, directness of the evidence, and precision of the estimate provided by the literature (Tables 48-51).
Table 48
Strength of evidence for effects of psychosocial interventions targeting parenting practices on parent-reported changes in disruptive behaviors in preschool children with DBD.
Table 49
Strength of evidence for effects of psychosocial interventions targeting parenting practices on parent-reported ratings of disruptive behaviors in school-age children with DBD.
Table 50
Strength of evidence for the effect of psychosocial interventions targeting parenting practices on parent-reported ratings of disruptive behaviors in teenage children with DBD.
Table 51
Strength of evidence for pharmacologic interventions.
Limitations
Limitations of This Review
This was a focused review on treatments for recognized disruptive behavior disorders at the individual level. Our focus was on treatments within a clinical setting or that might be a referral from a clinical setting. Therefore, we did not include ecologic approaches and psychosocial interventions that have been studied in settings such as juvenile delinquency settings and schools. These are important components of the overall therapeutic environment for disruptive behavior disorders and have been reviewed elsewhere.
We classified a heterogeneous group of interventions into the three broad categories of interventions that only include a child component, interventions that only include a parent component, or multi-component interventions. We defined multicomponent interventions as those that included two or more of a child component, parent component, or other component (e.g., teacher component, family together component). To account for the fact these treatment categories are broad, encompassing a range of specific interventions, each component was modeled as a random effect. This allowed for variation in treatment effect within each class. It is also worth noting that we classified PCIT as a multi-component intervention because, as its name suggests, the focus of the intervention is on the parent-child interaction and includes the parent and child engaged together in activities. Out of concern that PCIT in particular may also reasonably be classified as an intervention with only a parent component, we ran our quantitative model under both classifications (i.e., with PCIT categorized as a multi-component intervention and as an intervention with only a parent component). Classifying PCIT as an intervention with only a parent component did not significantly change our meta-analysis results, although point estimates of effect were nominally different.
We also excluded studies that focused on the treatment of other psychiatric conditions likely to have comorbid features of DBD. These would include, for example, ADHD, autism, and bipolar disorder. These are important and prevalent conditions and our review is intended to provide evidence on a very specific set of interventions in a defined group of participants. Clinical decisions need to be made with all of the available information, potentially from other reviews, particularly in complex clinical scenarios.
We did not limit inclusion to studies of individuals with a DSM diagnosis of DBD, but we did limit to those studies that provided some evidence that participants were beyond a validated clinical cutoff. Given the diversity of DBDs and a lack of consistent approach to assessing or reporting them, this is not a perfect approach. It is possible that some studies that did focus on DBDs were missed due to the reporting of the papers. We also focused on the outcomes that were by far the most commonly reported in the literature – e.g., parent reports of their child's disruptive behaviors – via measures assessing externalizing behaviors. Functional outcomes are also important. We recognize that parent reports of their child's disruptive behaviors are potentially biased, particularly when study designs did not include blinding. We also recognize that there are outcomes of interest such as emotional and psychological states beyond those that we specified in our protocol, but to widen the scope would have been infeasible for this review. Similarly, there is substantial overlap between several other psychological conditions and DBDs, including in particular ADHD. We did not include studies that focused primarily on treating ADHD, although some of these studies may also include evidence about disruptive behaviors as a component of ADHD. There are good reviews of ADHD treatment, including one by AHRQ,69 and we would point readers to those as additional information. We did include a number of studies that used traditional ADHD drugs but were focused on the disruptive behaviors themselves.
We were unable to review DBD interventions by etiology, although we understand that disruptive behaviors may stem from many causes (e.g. trauma), and these play into decisions about treatment and therapy.
Limitations of the Evidence Base
KQ1. Psychosocial Interventions
A number of methodological limitations exist in the literature base for child disruptive behavior disorders. First, identifying the target population is difficult. We included in our review both studies of children with a formal diagnosis of a disruptive behavior disorder and children without a formal diagnosis of a disruptive behavior disorder who scored above a clinical cutoff on a well-validated measure of child disruptive behaviors, but lack of detail in reporting by authors makes it challenging to fully and accurately characterize the populations in the studies.
Second, although most included studies were RCTs, overall the literature suffered from a lack of consistent and complete reporting. In particular, primary outcomes are rarely identified, and random sequence generation and allocation concealment rarely described. In addition, there was frequently no attempt to achieve blinding. Although there are well-recognized and valid reasons that achieving this level of control in the studies is challenging, if not impossible, it does bring some degree of potential risk of bias into the literature as a whole.
Third, the field lacks consensus on the most important outcomes. Few studies measure similar outcomes for synthesis. Methodologically, outcomes such as direct observation by a blinded and independent observer are arguably the most valid. However, direct observations can be expensive and are not always logistically feasible. From the perspective of patient-centered outcomes research, we believe that there is a strong argument in this literature to be made in favor of the importance of parent reported outcomes. However, most of the studied interventions included a parent component either alone or in combination with other components which introduces a potential risk of bias especially considering that blinding was not always feasible, and when parent reported outcomes were included multiple measures of similar constructs were used within and across studies. The reliance of the literature on parent reported outcomes and their potential for bias is a significant limitation of the evidence base.
Fourth, conflict of interest is a concern in this evidence base. Most of the studies evaluating a psychosocial intervention for a child disruptive behavior included in this review were conducted either by the developer of the intervention or by an “intellectual descendant” of the developer. Although it is understandable for this to be the case (much like it is understandable to see industry-sponsored clinical drug trials), the strength of the evidence for this body of literature would be strengthened with more studies independently evaluating the interventions.
Fifth, there are few direct comparisons of individual interventions. Specific interventions were most often compared to a waitlist control group or treatment as usual (variably described). When comparisons of active treatments were included, it was most often a comparison of different versions of a specific intervention. Further, results from mixed models are not always presented in a straightforward manner, making it very difficult to tease out effects of specific treatment approaches.
KQ2. Pharmacologic Interventions
There were surprisingly few studies focused on treating disruptive behaviors with pharmacologic interventions, which reflects the fact that these drugs are frequently used off label and without a research basis for their use in this particular set of disorders. Indications for the drugs reviewed here include a range of conditions, including but not limited to ADHD, schizophrenia, bipolar disorder, and seizures (complete list in Appendix I). As such, many of the studies include mixed populations and report outcomes of overlapping symptoms (e.g. of ADHD and DBD) making it difficult to discern the degree to which the mitigation of ADHD, for example, is in fact driving the results. Most of the studies in this section were small and larger studies are clearly needed.
Finally, it is a particular weakness that almost all studies were funded by the pharmaceutical company making the drug being studied. There is a clear need for replication and for independently funded studies.
KQ3 and KQ4. Combined Interventions
There were no studies to evaluate the efficacy of both behavioral and pharmacologic interventions compared to pharmacologic or behavioral interventions alone. Given that the clinical reality for many, if not most, families is that they use a multipronged approach for treatment of their children with DBDs, these studies are needed.
Future Research Needs
Research needs are both substantive and methodologic, and include both conduct and reporting of research. As noted above, randomization and allocation approaches were consistently not adequately described, and blinding was not attempted or addressed in much of the psychosocial literature. Future research should also clearly describe the duration of time from baseline to post-treatment and post-treatment to followup, and more clearly describe results from mixed models. Because the intervention developer is often the researcher, existing research must be replicated, as the lack of replication introduces the potential for a risk of bias analogous to that introduced by industry-sponsored trials of pharmaceutical interventions.
There is a need for specific, head-to-head comparisons of psychosocial interventions, evaluate the effectiveness of psychosocial interventions as compared to pharmacologic interventions (KQ3), and the effectiveness of combined psychosocial and pharmacologic intervention (KQ4). Additionally, prospective longitudinal studies examining implementation of these interventions in real-world community practice, including cultural adaptations, are also needed. End users urgently need this information to make informed decisions about which treatments to seek for their children. Clinicians need answers to these questions to decide which interventions to be trained to deliver and to recommend to their patients. Policymakers need this information to determine how to incentivize the provision of care for which there is the most evidence of effectiveness.
Future research should also clearly identify the target population and address the portability of studied interventions from predominantly university research clinics to real-world clinical settings. In the United States, disruptive behaviors are more prevalent among children receiving publicly funded care, and who are therefore likely to receive treatment in clinical settings such as community mental health centers. This group of young people may differ in important ways from the children receiving treatment in university-based research clinics. There is a growing body of literature about the challenges of implementing and disseminating best practices to real-world clinical settings with fidelity.245
Implications for Clinical and Policy Decisionmaking
Qualitatively and quantitatively, our review suggests that psychosocial interventions that include a parent component either alone or in combination with other components have the greatest probability of being most effective. This suggests that parents of children with disruptive behavior disorders should seek interventions that include a parent component. Clinicians providing care to this patient population should reconsider their current practices and clinicians referring families to specialty care should look to make referrals to clinicians whose interventions include a parent component. Researchers should consider more rigorously designed evaluations including of the potential harms of psychosocial interventions for this population, and policymakers and third party payers might consider writing clinical practice guidelines and reimbursement strategies that reflect this evidence.
There is less consistent evidence from the pharmacologic literature, but moderate SOE available for the use of antipsychotics and nonstimulant drugs. Parents of children with disruptive behaviors may, in consultation with their healthcare providers, want to consider the potential benefits of these pharmacologic options in the context of what is known about their risks. Many if not most clinicians providing pharmacologic care to this patient population are likely already aware of the potential benefits and harms of associated with use of these medications. Researchers may see potential for additional research on the effectiveness of these medications for this patient population. The implications for policymakers and third party payers are somewhat less clear.
Although we know from studies of other childhood disorders such as depression that combined psychosocial and pharmacologic intervention has benefits over either intervention alone,246 there is currently insufficient evidence to make similar conclusions for the treatment of children with disruptive behaviors. In reality, families of children with DBDs and the clinicians working with them are likely facing an array of treatment approaches to combat a complex set of symptoms or expressions of psychiatric conditions. This report should be assessed within the context of other reviews and primary literature. It provides evidence for one piece of a complex puzzle.
Conclusions
This review suggests that psychosocial interventions for children with disruptive behavior disorders that are multicomponent interventions or interventions that include only a parent component are likely to be more effective at reducing problem behaviors than psychosocial interventions that include only a child component or treatment as usual. As defined in this study, all multicomponent interventions included a parent component. Thus, it seems likely that a parent component is important. There are very few studies directly supporting the effectiveness of pharmacologic interventions for children with disruptive behavior disorders, but small studies of antipsychotics and stimulants report positive effects in the very short term. There are no studies examining the effectiveness of these interventions in combination with one another. The most commonly reported outcomes are parent-reported outcomes. Long-term and functional outcomes were not consistently reported. There was variability in the duration of long-term followup and functional outcomes reported. To date, treatment research is almost exclusively supported by the pharmaceutical industry. Given the prevalence of DBDs and the need for high quality data to inform clinical practice, more long-term studies are needed as are studies aimed at treating DBD separate from comorbid ADHD.
- Discussion - Psychosocial and Pharmacologic Interventions for Disruptive Behavio...Discussion - Psychosocial and Pharmacologic Interventions for Disruptive Behavior in Children and Adolescents
- Background - Psychosocial and Pharmacologic Interventions for Disruptive Behavio...Background - Psychosocial and Pharmacologic Interventions for Disruptive Behavior in Children and Adolescents
- Adjuvant Treatment for Phenylketonuria (PKU)Adjuvant Treatment for Phenylketonuria (PKU)
- Adjuvant Treatment for Phenylketonuria: Future Research NeedsAdjuvant Treatment for Phenylketonuria: Future Research Needs
- Maternal-Fetal Surgical ProceduresMaternal-Fetal Surgical Procedures
Your browsing activity is empty.
Activity recording is turned off.
See more...