Effectiveness of educational setting-based interventions for preventing conduct disorder

Systematic review results

Studies included in the review

The overall PRISMA flow diagram for the project is reported in Figure 2. Twenty-seven papers relating to five studies met our eligibility criteria for inclusion in the conduct disorder review. Included studies were published between 1999 and 2018, and randomised between 225 and 891 participants (median 245 participants). All studies were cluster randomised, of which four reported results from appropriate models incorporating the multilevel nature of the data.

One study was classified as universal243 and four were classified as targeted244–247 (all indicated). All studies were conducted in primary school settings, with age at baseline ranging from 4.2 to 7.91 years. The majority of study participants were boys. In the four indicated studies, the proportion of boys ranged from 69% to 74% of participants. Four studies were conducted in HICs,243,244,246,247 and one was conducted in a MIC.245 Of the studies conducted in HICs, all were conducted in lower-income settings, as specified by the trial authors.

Owing to intervention complexity and the flexible nature of the intervention implementation, we were not able to calculate an average ‘dose’, as was done for anxiety and depression outcomes.

Risk-of-bias assessments are reported in Figure 15 by individual study. Three studies were judged as having unclear risk of bias for random sequence generation and allocation concealment.243,244,247 All studies were judged to be at high risk of bias for participant and outcome assessor blinding. Prospective trial registrations were available for one study.245 For cluster randomised trials we also considered how recruitment, randomisation and analysis were conducted under the Cochrane Risk of Bias tool heading of ‘other bias’. Four studies were judged to be at high or unclear risk for ‘other bias’.

FIGURE 15 Conduct disorder risk-of-bias assessments by domain and study. +, Low risk of bias; –, high risk of bias; ?, unclear risk of bias. (Figure not reproduced here; see the original article.)

Owing to the diversity of interventions, outcome measures and time points reported by the studies, results are reported narratively by trial, and by date of publication. Owing to this diversity, intervention components are also described narratively, by study, where these could be identified from trial reports. Unless otherwise stated, statistical summaries are reported as described in the original publications.

Fast Track: the Conduct Problems Prevention Research Group247

The Fast Track trial247 was a cluster randomised trial of 54 primary schools in the USA. Beginning in 1991, the trial followed participants for > 20 years. The indicated prevention intervention was delivered over 10 years, targeting children in primary school settings with early-starting disruptive behaviour. The aim of the intervention was to prevent long-term severe and chronic conduct problems, including psychological disorder. The study compared a multicomponent intervention (n = 445) with a no-intervention control (n = 446). Children who were within the top 10% of a combined parent–teacher screen for externalising behaviour problems were invited to take part. The mean age of child participants was 6.5 years (SD 0.48 years); 69% were boys, 51% were African American and 47% were European American. The schools were in areas of high crime and poverty.

Seven intervention components were implemented, the content of which changed each year to be developmentally appropriate for the children and based on the families’ needs. There was insufficient detail provided in the reports to describe the exact nature of these changes. The intervention described subsequently is from the primary school phase of the project (i.e. grades 1–3, ages 6–9 years).

The school-based component was delivered to whole classrooms, and not just to the students enrolled in the Fast Track trial. The PATHS^® (Promoting Alternative THinking Strategies) intervention was delivered by classroom teachers across two or three lessons per week across the whole school year. PATHS is a social and emotional learning intervention that focuses on children managing and understanding their behaviour and emotions and includes social and problem-solving skills training. Fidelity was checked by Fast Track educational co-ordinators via weekly classroom and teacher visits. Educational co-ordinators were usually ex-teachers.

There were five parent and/or child components: parent groups, child social skills training groups, parent–child sharing time, child–peer pairing and academic tutoring. Parent groups, child social skills training groups and parent–child sharing time ran during a weekly 2-hour enrichment programme held at school but outside school hours. A total of 22 sessions was offered. Average attendance was 78% for the child group and 71% for the parent group. The child groups focused on reviewing and practising skills in emotional understanding and communication, friendship-building, self-control and social problem-solving. The parent groups focused on parenting skills, self-control and building positive family–school relationships.

Individual support was provided to children and parents by family co-ordinators; this formed the home visiting component. Visits focused on problem-solving and encouraging parental empowerment and self-efficacy. Family co-ordinators typically had advanced degrees in counselling or social work. Home visits were typically every other week across the first year of the intervention. Children were also provided with academic tutoring three times each week for 30-minute sessions during school hours, delivered by ‘paraprofessional tutors’.

Seventy-two per cent of participants received > 50% of each of the following intervention components: parent group, child social skills group, peer pairing and tutoring. Overall, 81% of participants received at least 50% of the recommended number of home visits (i.e. at least six visits). All programme staff attended a 3-day cross-site workshop. Fidelity was monitored using intervention manuals, weekly supervisory telephone calls and weekly staff meetings.

Multiple assessment scales were administered; primary outcomes were not specified. Appropriate multilevel models were conducted for the end of year 1 analyses, but not for later time points. At the end of grade 1 (i.e. 1 year into the intervention), the following effect sizes (Cohen’s d) were reported by the Conduct Problems Prevention Research Group,247 as computed from the F-value and degrees of freedom. Parent ratings of child behaviour change (Cohen’s d 0.50, 95% CI 0.25 to 0.76) and teacher ratings of child behaviour change (Cohen’s d 0.53, 95% CI 0.28 to 0.79) suggested a beneficial effect of the intervention, compared with the control. However for externalising behaviours, parent ratings on the Child Behaviour Checklist (CBCL) (Cohen’s d 0.04, 95% CI –0.15 to 0.26) and teacher ratings on the Teacher Report Form (Cohen’s d 0.02, 95% CI –0.19 to 0.24) suggested no evidence of a difference between intervention and control groups at the end of grade 1. At the end of grade 3, parent ratings of child behaviour change (β 0.18, 95% CI 0.04 to 0.32) and teacher ratings of child behaviour change (β 0.24, 95% CI 0.12 to 0.35) also suggested a beneficial effect for the intervention, compared with the control. However, the parent-rated CBCL was not reported and the teacher-rated Teacher Report Form (β –0.05, 95% CI –1.42 to 1.32) externalising problems scales suggested no evidence of a difference between the intervention and control groups at the end of grade 3.

Three further measures were reported at the end of grade 3. Academic progress in reading (β 0.06, 95% CI –0.06 to 0.17) and mathematics (β 0.08, 95% CI –1.42 to 1.32), and clinical diagnosis of oppositional defiant disorder (ODD) or conduct disorder (β 0.02, 95% CI –0.04 to 0.08). There was no evidence to suggest an intervention effect on these outcomes. Note that the 95% CIs for these effect sizes and regression coefficients are an approximation248 based on data reported by the Conduct Problems Prevention Research Group across two papers.247,249

In grades 7 and 8, there was no evidence of a difference between intervention and control groups for the parent-rated CBCL (β 1.50, 95% CI –0.15 to 3.15) or the teacher-rated Teacher Report Form (β –0.33, 95% CI –2.45 to 1.79) externalising problems scales. Hyperactivity was also reported and was reduced in the intervention group, compared with the control group.

An independent analysis of the data suggests that there was also little evidence of an effect on mental health outcomes during ‘the high school years’ (i.e. between 9 and 13 years of follow-up).250 However, the Fast Track trial team conducted a further follow-up, 20 years from baseline (when participants were aged 25 years). On the basis of self-reported outcomes, there was strong evidence for a beneficial effect of the intervention for a composite outcome of any externalising, internalising or substance use problem (OR 0.59, 95% CI 0.43 to 0.81). The authors of the paper state that this is equivalent to a number needed to treat of 8. For diagnosis of antisocial personality disorder, there was a benefit of the intervention, relative to the control (OR 0.60, 95% CI 0.39 to 0.93), but not for attention deficit hyperactivity disorder (OR 0.65, 95% CI 0.39 to 1.08). There was no evidence to suggest that the intervention prevented diagnoses of anxiety (OR 0.79, 95% CI 0.47 to 1.33) or depression (OR 0.68, 95% CI 0.42 to 1.08). Analyses were based on the intention-to-treat principle and appropriately accounted for clustering.

Substance use outcomes included alcohol abuse (OR 0.69, 95% CI 0.48 to 0.99), binge drinking (OR 0.75, 95% CI 0.55 to 1.01), heavy marijuana use (OR 0.76, 95% CI 0.45 to 1.30) and serious substance use (OR 0.58, 95% CI 0.36 to 0.92). There was no evidence of a difference between intervention and control groups for having graduated from high school (OR 0.93, 95% CI 0.68 to 1.27) or currently being in full-time education or employment (OR 0.84, 95% CI 0.60 to 1.18).

Early Risers Skills for Success: August et al.244

The Early Risers ‘Skills for Success’ study244 was a cluster randomised trial for the prevention of serious behaviour problems, as indicated by aggressive–disruptive behaviour. Ten schools were randomised to a multicomponent intervention and 10 were randomised to no intervention. Participants were screened prior to trial entry, to ensure that they were at high risk of a serious conduct problem, as assessed by a T-score of ≥ 58 on the CBCL aggression subscale (which was teacher rated). A total of 245 children were included in the trial: 124 in intervention schools and 121 in control schools. The mean age of participants at baseline was 6.6 years (SD 0.6 years) and 69% were male. Schools were in semirural midwestern USA, in low to low–middle socioeconomic areas. There have been multiple follow-ups of the participants, the latest being 10 years from the start of intervention (i.e. 5 years post intervention).

The intervention was multicomponent and multiphase. Beginning in the summer preceding grade 1 (i.e. year 1 of the intervention) and continuing over the next 2 years (i.e. 3 years in total), students attended a 6-week summer school, equivalent to 432 hours of programme content. Participants received academic content, enrichment activities and small-group social skills training. Peer mentoring was also used. It is not clear who delivered the summer school component of the intervention. In years 4 and 5, ‘booster’ summer camps were offered, each of 6 days’ duration.

The second intervention component was a monitoring and mentoring programme overseen by ‘family advocates’, who provided weekly support to the participants and teachers in their regular school setting. Support provided was flexible, depending on student needs. The third component was a biweekly family programme, based on the Incredible Years intervention, delivered concurrently, but separately, to parents and children. The Incredible Years programme is described elsewhere as being based on CBT principles, but was not described as such here.251 Twenty-nine sessions, over 3 years, were offered, and were equivalent to 58 hours in total. However, attendance was low [mean 39% (SD 29%)]. In years 4 and 5, six family sessions were offered. In addition to the previously mentioned core intervention components, a personalised home visitation component was available on the basis of individual need.

Approximately 60% participated across all three intervention components in the first 3 years, and 67% participated in three or more components offered during the booster phase. The intervention was manualised, and training was provided to all staff. Intervention fidelity was measured via log books and checked during unannounced visits by ‘fidelity technicians’.

At the end of the main intervention (year 3), externalising outcomes were reported as a composite across a maximum of four measurement scales for outcomes of (1) aggression, (2) hyperactivity and (3) impulsivity. The scales were the Teacher Observation of Classroom Adaptation, the Parent Observation of Classroom Adaptation, the Behavior Assessment System for Children (BASC) – Teacher Rating Scale (BASC-TRS) and the BASC – Parent Rating Scale (BASC-PRS). Teacher-reported academic achievement was also a composite outcome combined across four separate scales (the broad reading and applied problems composite scores from the Woodcock–Johnson Tests of Achievement-Revised, the Learning Problems scale of the BASC-TRS, and the Cognitive Competence scale of the Teacher’s Scale of Child’s Actual Competence and Social Acceptance). An intention-to-treat, three-level model was appropriately conducted, given the clustered nature of the data; however, the school level was retained for the aggression analysis only.

At the end of year 3, intervention participants (n = 199) showed a greater improvement in academic achievement than control participants. However, there was no evidence of an effect for aggression, hyperactivity or impulsivity, and quantitative results were not reported. At the end of year 6 (n = 151), outcomes included self-reported conduct disorder symptoms and diagnosis, ODD symptoms and diagnosis, and self-reported drug use. Based on the means and SDs reported by the authors, we calculated that there was no evidence that the intervention prevented self-reported conduct disorder symptoms (mean difference –0.20, 95% CI –0.91 to 0.51). However, there was evidence to suggest that, for ODD symptoms, the intervention was beneficial (mean difference –1.53, 95% CI –2.58 to –0.48). The authors reported that the ODD effect size was 0.47 (p ≤ 0.01). The effect size for conduct disorder symptoms was not reported. There was no evidence that conduct disorder or ODD diagnoses differed across the intervention and control groups. There was no evidence of an effect for tobacco or alcohol use.

At 10 years from baseline (n = 129), self-reported conduct disorder and ODD were reported. The authors report that the Early Risers intervention was associated with fewer conduct disorder and ODD symptoms in late high school. The estimated mean number of conduct disorder symptoms per participant was 1.81 lower (95% CI 0.34 to 3.30) in the programme group than in the control group and the mean number of ODD symptoms was 1.56 lower (95% CI 0.47 to 2.63).

Incredible Years teacher programme: Baker-Henningham et al.245

Baker-Henningham et al.245 reported a cluster randomised trial for the indicated prevention of disruptive and aggressive/destructive behaviours. Twenty-four preschools were included and were randomised to either the Incredible Years intervention group (n = 12) or a control group (n = 12). It was not clear from the paper what intervention the control group received, although the trial registration (ISRCTN 35476268) suggests that it was information only. Children were selected for inclusion by a teacher-rated scale based on the ICD-10 criteria for conduct disorder. Three children with the highest scores in each class were enrolled in the study. A total of 225 children aged 3–6 years were included (113 in the intervention group and 112 in the control group). The mean age of participants was 4.2 years (SD 0.9 years), and 69% of participants were boys. Schools were in Kingston, Jamaica, and in ‘disadvantaged’ inner-city areas with high levels of community violence.

The intervention evaluated was the Incredible Years Teacher Training programme, tailored to a Jamaican context. Few details on the intervention were provided in the paper, but it included collaborative and experiential learning; individual goal-setting and self-monitoring; building teachers’ self-efficacy; a focus on teachers’ cognitions, behaviour and emotions; and emphasis on teachers’ ability to generalise the skills learnt. The Incredible Years programme has been described elsewhere as being based on CBT principles, but we note that it was not described as such here.251 The intervention lasted for 6 months. To ensure fidelity, intervention teachers attended eight full-day training workshops over the period of the intervention and received in-class support from a psychology graduate with previous experience in Incredible Years. The authors reported that teachers attended a median of eight workshops (range 2–8 workshops), with 95% attending at least six workshops; 89% of teachers received all four in-class consultations.

The primary outcome was directly observed in-class child behaviour and reported as frequency of aggressive/destructive behaviours (as defined by a study manual). Secondary outcomes of interest were teacher and parent reports of child behaviour. Teacher-reported child conduct problems were measured using the Sutter–Eyberg Student Behavior Inventory™ (SESBI), and the SDQ was used to measure behaviour difficulties and prosocial skills. Parent-reported child conduct problems were measured using the Eyberg Child Behavior Inventory™ (ECBI), and the SDQ was used to measure behaviour difficulties and prosocial skills. Hyperactivity and attention difficulties were also measured using Conners Global Index. These scales formed a composite outcome of ‘behavioural difficulties’. Child school attendance was taken from school records.

Statistical analyses appropriately accounted for the clustered nature of the data. For the author-reported primary outcome, the intervention reduced the number of directly observed conduct problems (effect size 0.42, 95% CI 0.12 to 0.71). For author-reported secondary outcomes, the intervention reduced the number of teacher-reported behavioural difficulties (effect size 0.47, 95% CI 0.18 to 0.76) and parent-reported behavioural difficulties (effect size 0.22, 95% CI 0.03 to 0.42), relative to the control group. School attendance was also higher in the intervention group than in the control group (effect size 0.30, 95% CI 0.05 to 0.55). Children in the intervention group were less likely to be rated in the clinical range for conduct disorder by teachers (OR 0.31, 95% CI 0.11 to 0.92) than by parents (OR 0.56, 95% CI 0.27 to 1.16).

Multisystemic early intervention: Havighurst et al.246

Havighurst et al.246 reported a cluster RCT of a multisystemic indicated intervention for primary school-aged children at risk of presenting with conduct disorder. Thirty-three schools in lower socioeconomic areas of Victoria, Australia, were included and were randomised to either the intervention (n = 14) or to a waiting list group (n = 19). Children between the ages of 5 and 9 years were eligible, and those scoring in the top 8% on a joint parent–teacher screen for behaviour problems (using the Conduct Problems Risk Screen) were invited to participate in the study. A total of 113 child participants were randomised to the intervention, and 118 were randomised to the waiting list control group. The mean age was 7.05 years (SD 1.06 years) and 74% of participants were boys.

The intervention contained separate parent, child and school components. The parent component was delivered across eight 2-hour sessions and focused on emotion socialisation coaching, whereby parents learn to respond positively to their children’s emotions. Average parent attendance was six sessions, with 78% of parents attending five or more. The intervention was delivered by clinical or educational psychologists, social workers, speech and language therapists, or occupational therapists. To ensure fidelity, intervention facilitators attended 2-day training, followed an intervention manual and completed weekly checklists. Intervention fidelity was rated as consistently high, with 100% of the foundation skills delivered and 78% of the optional skills delivered.

The child component focused on skills in emotional competence, de-escalation of anger and social problem-solving. Eight sessions were delivered to small groups during school time. Groups were facilitated by two professionals: an intervention clinician and a member of school staff (often a school psychologist or teacher). Average attendance was 7.3 sessions, with 84 children (92.3%) attending at least six sessions. To ensure fidelity, intervention facilitators attended a half-day training, followed a structured intervention manual and completed weekly checklists. One hundred per cent of the child programme content was covered in all groups.

In addition to the parent and child components, schools were also offered the choice between two universal interventions. Seven schools (32 children) received the PATHS intervention, and an additional seven schools received a Professional Learning Package (59 children). Fidelity was not measured.

Follow-up assessments were conducted 10 months post baseline. The following effect estimates are based on adjusted means as reported by the authors. Primary outcomes of interest included parent-reported child behaviour, as measured by the Eyberg Child Behavior Inventory (ODD, conduct disorder and hyperactivity subscales), and teacher-reported child behaviour, as measured by the total SDQ score. Statistical analyses appropriately accounted for the clustered nature of the data and were based on the intention-to-treat principle. For parent-reported child behaviour, there was a beneficial effect of the intervention for conduct disorder (mean difference –2.94, 95% CI –3.43 to –2.45), ODD (mean difference –4.75, 95% CI –5.51 to –3.99) and hyperactivity (mean difference –3.47, 95% CI –3.89 to –3.05), relative to a waiting list control. For teacher-reported child behaviour, the intervention reduced the total SDQ score, relative to the control group (mean difference –1.56, 95% CI –1.80 to –1.32).

Kyranides et al.243

Kyranides et al.243 reported a small (three schools) cluster RCT of a universal intervention for preventing conduct disorder and callous unemotional traits among children between the ages of 7 and 9 years. Three schools in areas of low SES were randomised to either a skills-building intervention (n = 1) or usual curriculum control (n = 2). Ninety-four children were allocated to the intervention and 210 to the control. The mean age was 7.9 years (SD 0.74 years) and 51% of participants were female.

The (unnamed) intervention was 8 weeks long, with one 45-minute session delivered each week during school hours. The intervention was based on CBT with an added emotional component and aimed to increase children’s awareness of their own and others’ emotions, teach self-control and emotion regulation, promote a positive self-concept, improve social skills and peer relations, and develop problem-solving and communication skills. The intervention was delivered to whole classes by PhD (Doctor of Philosophy) students with master’s degrees in school psychology. An intervention manual was provided, and fidelity was monitored by the research supervisor.

Primary outcomes of interest to this review are the parent-reported Checkmate Child Symptom Inventory-4 to assess symptoms of conduct disorder and the Antisocial Process Screening Device to assess impulsivity. Analyses did not consider clustering. Immediately post intervention (mean difference –0.67, 95% CI –1.30 to –0.04) and at 9 months post intervention (mean difference –0.93, 95% CI –1.32 to –0.55), there was evidence to suggest that conduct disorder symptoms were reduced in the intervention group, relative to the control group. Impulsivity was reported for the post-intervention time point only: the result suggested that there was a beneficial effect of the intervention, relative to the usual curriculum control (mean difference –1.03, 95% CI –1.72 to –0.34).

Summary

Only two studies clearly specified a post-intervention time point.243,245 Results for all studies at the time point most closely approximating post intervention are summarised in Table 17. All studies reported parent- or teacher-reported outcomes. Only one study reported the primary review outcome of conduct disorder symptoms;243 the remainder reported externalising behaviours or behaviour difficulties.

Two studies243,245 implemented school-only interventions, one of which described the intervention as based on CBT principles.243 For the primary time point of post intervention, both studies reported evidence of a beneficial effect of intervention. One study245 was judged to be at low risk of bias for allocation concealment and randomised sequence generation; the other243 was judged to be at unclear risk of bias for both domains.

Three studies implemented multisystemic, multicomponent and multiphase ‘packages’ of interventions.244,246,247 Based on the components reported, it is possible that the child-focused school-based components of these interventions were informed by cognitive–behavioural principles. However, none of the authors identified them as CBT-based interventions and their results were reported at the combined ‘package’ level. The studies also reported different follow-up time points. At the (presumed) post-intervention time point, Havighurst et al.246 reported strong evidence of a beneficial effect of the intervention for parent-reported child behaviour, relative to a waiting list control. This study was rated as having a low risk of bias for randomised sequence generation, but an unclear risk of bias for allocation concealment. No further follow-up time points were reported.

Two studies244,247 reported interventions that, in addition to being multisystemic, were implemented in stages over several years, so it is difficult to discern a ‘post-intervention’ time point. The Fast Track247 study did not clearly specify a primary outcome, and the outcome measures reported vary at each time point. At 1 year and 3 years from baseline, the authors reported a beneficial effect of intervention for child behaviour change, but not for externalising behaviours. At 7 years’ follow-up, the authors reported no evidence of an effect of the intervention for reducing externalising behaviours. Finally, at 20 years’ follow-up, the authors reported a composite internalising, externalising and substance use self-reported outcome and concluded that there was strong evidence of an effect of the intervention. The Fast Track247 study rated as having an unclear risk of bias for both randomised sequence generation and allocation concealment.

At 3 years from baseline, August et al.244 reported that there was no evidence that the Early Risers intervention reduced teacher-/parent-reported externalising behaviours. At 6 years from baseline, there was no evidence that the intervention prevented self-reported conduct disorder symptoms, although there was a beneficial effect for self-reported ODD symptoms. At 10 years from baseline, the authors reported that there was evidence that the intervention was associated with fewer self-reported conduct disorder symptoms, relative to no intervention. The study was rated as having an unclear risk of bias for both randomised sequence generation and allocation concealment.