The Pharmacological Management of Oppositional Behaviour, Conduct Problems, and Aggression in Children and Adolescents With Attention-Deficit Hyperactivity Disorder, Oppositional Defiant Disorder, and Conduct Disorder: A Systematic Review and Meta-Analysi (2024)

PMCID: PMC4344946

Tamara Pringsheim, MD, MSc,¹ Lauren Hirsch, BSc (MSc Candidate),² David Gardner, PharmD, MSc,³ and Daniel A Gorman, MD⁴

Author information Article notes Copyright and License information PMC Disclaimer

Objective:

Children with attention-deficit hyperactivity disorder (ADHD) may have oppositional behaviour, conduct problems, and aggression. These symptoms vary in severity, and may be related to a comorbid diagnosis of oppositional defiant disorder (ODD) or conduct disorder (CD). Critical evaluation of the efficacy of ADHD medications may guide the clinician regarding the usefulness of medications for these symptoms.

Method:

We performed a systematic review and meta-analysis of psychostimulants, alpha-2 agonists, and atomoxetine for oppositional behaviour, conduct problems, and aggression in youth with ADHD, ODD, and CD. The quality of evidence for medications was rated using the Grading of Recommendations Assessment, Development and Evaluation approach.

Results:

Two systematic reviews and 20 randomized controlled trials were included. There is high-quality evidence that psychostimulants have a moderate-to-large effect on oppositional behaviour, conduct problems, and aggression in youth with ADHD, with and without ODD or CD. There is very-low-quality evidence that clonidine has a small effect on oppositional behaviour and conduct problems in youth with ADHD, with and without ODD or CD. There is moderate-quality evidence that guanfacine has a small-to-moderate effect on oppositional behaviour in youth with ADHD, with and without ODD. There is high-quality evidence that atomoxetine has a small effect on oppositional behaviour in youth with ADHD, with and without ODD or CD.

Conclusions:

Evidence indicates that psychostimulants, alpha-2 agonists, and atomoxetine can be beneficial for disruptive and aggressive behaviours in addition to core ADHD symptoms; however, psychostimulants generally provide the most benefit.

Objectif :

Les enfants souffrant du trouble de déficit de l’attention avec hyperactivité (TDAH) peuvent avoir un comportement oppositionnel, des problèmes de conduite, et de l’agressivité. La gravité de ces symptômes peut varier, et ils peuvent être liés à un diagnostic comorbide de trouble oppositionnel avec provocation (TOP) ou de trouble des conduites (TC). L’évaluation critique de l’efficacité des médicaments du TDAH peut guider le clinicien à l’égard de l’utilité des médicaments pour ces symptômes.

Méthode :

Nous avons effectué une revue systématique et une méta-analyse des psychostimulants, des agonistes alpha-2 et de l’atomoxétine pour le comportement oppositionnel, les problèmes de conduite, et l’agressivité chez les adolescents souffrant de TDAH, de TOP et de TC. La qualité des données probantes pour les médicaments a été cotée à l’aide de l’approche de Classem*nt de l’analyse, de l’élaboration et de l’évaluation des recommandations.

Résultats :

Deux revues systématiques et 20 essais randomisés contrôlés ont été inclus. Des données probantes de qualité élevée indiquent que les psychostimulants ont un effet de modéré à grand sur le comportement oppositionnel, les problèmes de conduite, et l’agressivité chez les adolescents souffrant du TDAH, avec et sans TOP ou TC. Des données probantes de très faible qualité indiquent que la clonidine a un effet modeste sur le comportement oppositionnel et les problèmes de conduite chez les adolescents souffrant du TDAH, avec et sans TOP ou TC. Des données probantes de qualité modérée montrent que la guanfacine a un effet de modeste à modéré sur le comportement oppositionnel chez les adolescents souffrant de TDAH, avec et sans TOP. Des données probantes de qualité élevée indiquent que l’atomoxétine a un effet modeste sur le comportement oppositionnel chez les adolescents souffrant du TDAH, avec et sans TOP ou TC.

Conclusions :

Les données probantes indiquent que les psychostimulants, les agonistes alpha-2, et l’atomoxétine peuvent être bénéfiques pour les comportements perturbateurs et agressifs qui s’ajoutent aux symptômes de base du TDAH. Cependant, les psychostimulants sont généralement plus avantageux.

Clinical Implications

• Among the medications used for the treatment of ADHD, psychostimulants have the most evidence for efficacy in the treatment of oppositional behaviour, conduct problems, and aggression.

• There is evidence to support the use of guanfacine and atomoxetine for oppositional behaviour, though effect sizes are small to moderate.

• The effect of clonidine on oppositional behaviour and conduct problems may not be clinically significant.

Limitations

• There are a very limited number of studies of guanfacine and clonidine for the treatment of oppositional behaviour, conduct problems, and aggression.

Types of Studies

We included systematic reviews and RCTs, both parallel group and crossover designs. Studies had to include outcomes on oppositional behaviour, conduct problems, or aggression, and had to include a placebo phase or group. We searched for studies regardless of publication type.

Types of Participants

We included children and adolescents with a diagnosis of ADHD, ODD, or CD, made by an established classification system. Given the high rate of comorbidity between these disorders and both subaverage IQ and chronic tic disorders, we did not exclude studies that included children with these comorbidities.

Types of Interventions

1. Psychostimulants (methylphenidate and amphetamine) at any dosage or formulation (for example, short- or long-acting medications).

2. Alpha agonists (clonidine and guanfacine) at any dosage or formulation.

3. Atomoxetine at any dosage.

Types of Outcome Measures

1. Oppositional behaviour, conduct problems, or aggression as measured by validated clinician-, parent-, or teacher-reported scales.

2. Adverse effects, including physical, laboratory, or electrocardiogram abnormalities, and adverse event–related drop outs.

Electronic Searches

We searched the CENTRAL (Cochrane Central Register of Controlled Trials), MEDLINE, and PsycInfo databases in October 2013 (start date from database inception) using the strategies in online eAppendix 1. Searches were performed by medication class. Language restrictions were not imposed. Studies were not excluded based on publication status.

Searching Other Resources

We searched ClinicalTrials.gov database for unpublished studies, and checked the references of all review articles found in our searches.

Selection of Studies

Two authors independently reviewed titles and abstracts from the searches and selected potentially relevant studies. Full-text articles were read in detail to determine if they fulfilled inclusion criteria. Disagreements were resolved through discussion between the reviewers. We did not need to refer any disagreement to an independent arbiter.

Data Extraction and Management

Two authors independently extracted data from studies and entered them into a predesigned data extraction form. The following data were extracted and entered:

1. Study design.

2. Randomization method.

3. Allocation concealment.

4. Blinding of participants, personnel, and outcome assessors.

5. Inclusion and exclusion criteria.

6. Number of participants.

7. Age distribution.

8. Sex.

9. Loss to follow-up.

10. Premature discontinuation and reasons for such.

11. Outcomes.

12. Incomplete outcome data and selective outcome reporting.

13. Method of analysis.

14. Comparability of groups at baseline.

We compared extracted data to ensure accuracy, and we resolved discrepancies through discussion.

Assessment of Methodological Quality of Individually Included Studies and Systematic Reviews

Two authors independently assessed the risk of bias of each study according to predetermined criteria. For each criterion, included studies were given ratings of low risk of bias, high risk of bias, or uncertain risk of bias. Based on the combination of these individual factors, studies were given an overall rating of class I, class II, or class III for methodological quality. See online eAppendix 2 for the methodological quality rating system. Included systematic reviews were evaluated for methodological quality using the R-AMSTAR tool and scored up to 44 points.⁵

Measures of Treatment Effect

For continuous outcomes, we used SMDs between treatment and placebo. The SMD is used to pool outcomes measured on different scales across studies. Depending on the presentation of individual study results, SMDs were calculated based on end point or change scores. For binary outcomes, we used odds ratios. Intention-to-treat analyses were performed: all randomized participants were included in the analyses and retained in the group to which they were allocated.

Unit of Analysis Issues

For crossover trials, we analyzed first period data only, when reported. When this was not reported, we took all measurements from treatment periods and placebo periods and analyzed these as if the trial were a parallel group trial. In crossover trials reporting results from multiple dosages of methylphenidate, only the data on the highest-included dosage were used in the meta-analysis.

Dealing With Missing Data

When standard deviations were missing, they were calculated from given standard errors and group sample size. When both the standard deviation and standard error were not provided, standard deviations were imputed from a study using the same measurement instrument, treatment, and population type. This was required for one study.

Assessment of Heterogeneity

We assessed clinical heterogeneity by comparing trial design and the distribution of participant factors. We assessed statistical heterogeneity by examining the I² statistic, an approximate quantity that describes the proportion of variation in point estimates that is due to heterogeneity of studies rather than to sampling error. We performed a chi-square test of hom*ogeneity to determine the strength of evidence that heterogeneity is genuine.

Assessment of Reporting Biases

Given the limited number of RCTs, we did not formally assess for reporting bias. Both included systematic reviews (for psychostimulants and atomoxetine) had an adequate number of RCTs to perform tests of reporting bias, which are described in the Results.

Data Synthesis

We performed meta-analysis on the data using both random-effects and fixed-effects models, and compared the results of these 2 approaches to assess statistical heterogeneity. If the I² statistic was less than 40%, a fixed-effect model was used. Meta-analysis was performed on efficacy data only, as there was great variability in the reporting of adverse effects. Data on the type and frequency of adverse effects are described for each included study.

Overall Evidence Quality Rating for Each Medication

Each medication was given an overall quality of evidence rating based on the GRADE approach.⁶ The overall quality of evidence is rated as high, moderate, low, or very low, based on the confidence in the estimate of effect. With high-quality evidence, further research is very unlikely to change our confidence in the estimate of effect. With moderate-quality evidence, further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. With low-quality evidence, any estimate of effect is very uncertain, and further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.

We determined an initial rating of the quality of evidence for each medication based on the number and quality of the individual studies. We required multiple RCTs without serious limitations to make an initial rating of high-quality evidence. Multiple RCTs with some serious limitations resulted in an initial rating of moderate-quality evidence. An initial rating of low-quality evidence was given when RCTs had very serious limitations and inconsistent results. GRADE then specifies 5 reasons for rating down the quality of evidence (risk of bias, imprecision, inconsistency, indirectness, and publication bias) and 3 reasons for rating up the quality of evidence (a large magnitude of effect, a dose–response gradient, and a situation in which all plausible biases would serve to increase our confidence in the estimated effect). Each of these factors was evaluated for each medication, leading to a final determination of the quality of evidence, which are presented in the evidence profile table (Table 1).

Study Participants

Most studies included youth with ADHD with and without ODD or CD. All studies included more boys than girls.

Study Outcomes

Several different scales were used to measure oppositional behaviour, conduct problems, or aggression. Most studies used rating scales developed by Keith Conners, which have been extensively revised and updated over 4 decades. The version of the instrument used was largely dependent on the time the RCT was performed. Items on the Conners instruments are based on the Diagnostic and Statistical Manual of Mental Disorders criteria for ADHD, ODD, and CD. These scales are widely used, both in clinical practice and in research, with a large sample of normative data available for comparison. The scales have overall excellent reliability and validity, and are available in full-length, short, and index forms, both for parents and for teachers.⁷ Other rating scales used in the included studies were the Attention Deficit Disorder with/without Hyperactivity Comprehensive Teacher Rating Scale, the Disruptive Behaviour Scale, the Child Conflict Index, the Swanson Nolan and Pelham 4 ODD Subscale, the ADHD Symptom Checklist-4 Oppositional Defiant Subscale, the CGI—Improvement scale, and the CGI—Severity scale. Descriptions of each scale can be found in online eAppendix 4.

Psychostimulants

Our search found 1 systematic review of psychostimulants for the treatment of aggression in youth with ADHD; it included 28 RCTs published between 1970 to 2001.⁸ Therefore, we included this systematic review in our analysis and restricted our own systematic review to studies published from 2001 to 2013.

Connor et al⁸ performed a meta-analysis to evaluate the effect size for stimulants on overt and covert aggression in children with ADHD. This meta-analysis received an R-AMSTAR score of 29/44 points. The authors included 28 RCTs of methylphenidate (21 studies), amphetamine (5), and pemoline (2). These studies had a total of 683 subjects, with sample sizes ranging from 6 to 46 (mean 24, SD not provided in the article). The mean age of subjects was 9.7 years (range 7.7 to 14.4), and 75% had comorbid ODD or CD. The mean duration of treatment was 13 days (SD not provided in the article), and mean doses of methylphenidate, amphetamine, and pemoline were 22.18, 23.74, and 145.15 mg/day, respectively. Each study was given a quality rating out of 7, and the mean rating was 5.8 (range 5 to 7).

Overt aggression-related behaviours were defined as “aggression resulting in a direct confrontation with the environment”⁸,^{p 254} (for example, physical assault, verbal threats, oppositional and defiant behaviour, conduct problems, rage attacks, and irritability), whereas covert aggression-related behaviours were defined as “aggression that is furtive and hidden from the environment”^{p 254} (for example, cheating, lying, stealing, and fire-setting). Where possible, separate effect sizes (Cohen d) were calculated for clinician-, parent-, and teacher-rated overt and covert aggression. Overall effect sizes for overt and covert aggression were calculated as well. Adverse effects were not considered.

For overt aggression, weighted effect sizes comparing psychostimulant and placebo were significant (P < 0.001) based on clinician ratings in 18 studies (d = 0.77; 95% CI 0.63 to 0.88), parent ratings in 13 studies (d = 0.71; 95% CI 0.42 to 1.15), teacher ratings in 16 studies (d = 1.04; 95% CI 0.79 to 1.32), and overall ratings in all 28 studies (d = 0.84; 95% CI 0.70 to 1.02). Significant heterogeneity was detected among the effect sizes in each of these analyses, but calculation of a so-called file-drawer statistic suggested that publication bias was highly unlikely. In general, the presence of comorbid ODD or CD correlated negatively with effect size. Effect sizes based on overall ratings were comparable for studies of methylphenidate (d = 0.80) and amphetamine (d = 0.83), and dose was not significantly associated with effect size.

Seven of 28 studies reported effects of stimulant treatment on covert aggression. Clinician ratings were available in 6 studies (d = 0.81; 95% CI 0.20 to 1.43, P < 0.001), and overall ratings in all 7 studies were lower at d = 0.54 (95% CI 0.21 to 1.29). In each of these analyses, heterogeneity across results was observed. The file-drawer statistics suggested that publication bias was unlikely. Drug type and dose were not associated with effect size.

We found 12 RCTs comparing psychostimulants with placebo, published from the end date of the Connor systematic review⁸ to 2013, that met our inclusion criteria. Eleven evaluated methylphenidate^9–19 and 1 studied lisdexamphetamine²⁰ in children with ADHD. These 12 RCTs included a total of 1681 participants. Trial size ranged from 24 to 327 participants (mean 140, SD 113.1). All trials included children with ADHD comorbid with ODD or CD (44% to 93%). Both immediate-release and long-acting forms of methylphenidate were studied. Trial length ranged from 2 to 16 weeks. Two trials received a quality rating of class I, 4 were class II, and 6 were class III. Measures of aggression, oppositional behaviour, and conduct problems were included as a secondary outcome in all studies, and all studies reported a significant benefit with psychostimulants, compared with placebo. See online eAppendix 5 for study characteristics, quality ratings, results of individual RCTs, and details regarding adverse effects. See eAppendix 6 for a list of excluded studies.

Eight studies provided adequate data on end point scores to be included in a meta-analysis. The SMD between psychostimulants and placebo for teacher-rated oppositional behaviour, conduct problems, and aggression was 0.84 (95% CI 0.59 to 1.10; I² = 73%, P < 0.001), using a random effects model (Figure 1a). The SMD between psychostimulants and placebo for parent-rated oppositional behaviour, conduct problems, and aggression was 0.55 (95% CI 0.36 to 0.73; I² = 56%, P < 0.001), using a random-effects model (Figure 1b). The evidence profile for psychostimulants is presented in Table 1. There is high-quality evidence that psychostimulants have a moderate-to-large effect in the management of oppositional behaviour, conduct problems, and aggression in youth with ADHD, with or without ODD or CD.

Alpha Agonists

Eight studies met our inclusion criteria; 2 of guanfacine extended release, 1 of clonidine extended release, and 5 of clonidine immediate release. Six studies evaluated children with ADHD, 1 study evaluated children with ADHD and a chronic tic disorder, and 1 study evaluated children with mental retardation and hyperkinetic disorder. See online eAppendix 5 for study characteristics, quality ratings, results of individual RCTs, and details regarding adverse effects. See online eAppendix 6 for a list of excluded studies.

Atomoxetine

We found one systematic review and meta-analysis of atomoxetine, published in 2014, with ODD symptoms as an outcome (see online eAppendix 5 for details).²⁸ The authors included all published and unpublished data on atomoxetine from placebo-controlled, double-blind, randomized trials in youth with ADHD and average intelligence. Twenty-five trials were included, comprising 3928 youth. The majority of studies (23/25) were industry-funded, with a mean trial duration of 8.6 weeks (range 4 to 18 weeks), and a mean atomoxetine dose of 1.17 mg/kg/day. Comorbid ODD and (or) CD was present in 44.6% of study participants. The results for ODD outcomes are based on 15 comparisons involving 1907 patients. The overall effect size for ODD symptoms was 0.33 (95% CI 0.24 to 0.43; I² = 0%, P < 0.001). The authors found that the proportion of subjects with a diagnosis of ODD did not significantly influence the effect size for ODD symptoms (P = 0.41). Funnel plots were inspected to assess for publication bias for all outcomes; no indication of publication bias favouring atomoxetine was identified.

Ratings of trial methodological quality were not performed by the authors of the systematic review.²⁸ Our own review of methodological quality of the included studies found 4 studies rated class I, 5 studies rated class II, and 2 studies rated class III. The evidence profile for atomoxetine is presented in Table 1. Overall, there is high-quality evidence that atomoxetine has a small effect on oppositional defiant behaviour in youth with ADHD, with and without ODD or CD.

Click here to view.^{(97K, pdf)}

Click here to view.^{(76K, pdf)}

Click here to view.^{(112K, pdf)}

Click here to view.^{(69K, pdf)}

Click here to view.^{(172K, pdf)}

Click here to view.^{(143K, pdf)}