Your Choice: Randomised Controlled Trial of a CBT Informed Violence Reduction Programme

Your Choice: a programme designed and developed by the London Violence Reduction Unit in collaboration with the Association of London Directors of Children’s Services and the Children and Adolescent Mental Health Service. The intervention is targeted at young people, aged 11-17, who are judged (by their local education, social care, justice and/or medical systems) to be at medium to high risk of harm or of engaging in violent and anti-social behaviour, and who are eligible to receive publicly-provided assistance in dealing with these problems. The intervention has three main components:

1. A 1-week intensive training of youth workers (who are social workers in direct contact with young people in the target group) on techniques and tools used in Cognitive Behaviour Therapy (CBT). The training is delivered by lead practitioners.
2. An intensive programme of three weekly meetings between each young person enrolled in the treatment and their trained youth worker, for the duration of 12 weeks.
3. Regular clinical supervision of trained youth workers, delivered by the trainer therapist individually or in groups, to revise the cases of each young person participating in Your Choice and the practical delivery of CBT techniques.

2022-07-01

2023-08-31

Emotional and behavioural difficulties and pro-sociality assessed using the Strengths and Difficulties Questionnaire (SDQ).

The Strengths and Difficulties Questionnaire (SDQ) is is a well validated and widely used tool to measure psychological strengths and difficulties (Goodman and Goodman, 2009). Specifically, the questionnaire measures emotional difficulties, conduct problems, hyperactivity, peer relationships and pro-social behaviour. It is self-reported for young people aged 11 to 17. The SDQ version used at endline in our pilot study also includes a follow-up questionnaire especially designed to detect changes in psychological attributes in response to interventions.
The questionnaire is short, making it an ideal tool to apply in contexts where young people may be unsettled, anxious, or find it difficult to concentrate for long periods of time. Moreover, it focuses on detecting psychological difficulties that may be particularly common among the target population of young people, and which are likely directly related to their vulnerabilities and behaviour. Early signs of the impact of Your Choice are, therefore, most likely detectable on follow-up SDQ outcomes.

Engagement and exposure to crime and violence, as measured by a scale co-produced by the evaluation and project teams administered to both the young person and their lead practitioner.
Social connectedness, measured by a subscale of the the Student Resilience Survey.
Mental Well-Being, measured by The Short Warwick–Edinburgh Mental Well-being Scale (SWEMWBS).
Self-efficacy, measured by the New General Self-Efficacy Scale.
Self-regulation, measured by the Trait Emotional Intelligence Questionnaire – Adolescent Short Form (teique-asf) – self-regulation subscale.

Local Authorities (LAs), which are the local government units in the UK, deliver a bundle of supporting services to young people, and particularly to those in the target group for Your Choice. London has 32 LAs, and they have all been commissioned by the London Violence Reduction Unit to deliver Your Choice by integrating the new programme within the range of other services they normally provide to our target population of youth. The delivery of these services is implemented by teams of 4-5 youth workers, organised at the LA level by intervention areas (e.g. Youth Offending Service, Gangs and Exploitation, Adolescent Safeguarding, etc). There are usually various teams in the same intervention area within each LA.

Randomization happens within LA, at the team level. London LAs provided us with the full list of teams that are in direct contact with young people in Your Choice target group. They also identified the full list of teams that are currently available to be trained in CBT techniques. All LAs named at least two teams that are ready for training. From this list of eligible teams, the evaluators will randomly select one or two teams per LA. Wherever possible, teams will be randomly selected within intervention area. Selected teams will be trained; randomized-out teams will form the control group.

Our research design relies on the fact that LA youth services operate at capacity. This implies that each time a new young person joins the services, the team that takes on the case will be the one in the relevant set of services that has fewer cases in their stock. This suggests that the assignment of young people to teams, conditional on Local Authority and service type, is as good as random – an assumption we will carefully test using the wealth of information that will be gathered about the participants during the pilot.

Randomization done in the office by a computer.

Teams of youth practitioners (operating each within a particular youth service and local authority). Wherever possible, treated and control teams will be selected from the same youth service within each local authority.

Yes

We aim to have 32 youth practitioners teams to be selected (one in each local authority in London). At least one control team will be available in each local authority.

Based on the results of an earlier survey to local authorities, we expect treated teams to enrol at least 4 new children in Your Choice each month.

If the pilot runs for two months, we expect to have 4 x 2 x 32 = 256 young people assigned to treated teams of youth practitioners. At least the same number of young people will be assigned to control teams.

The table shows the minimum detectable effect sizes (MDE) under our baseline assumption of 32 treated teams (and the same number of control teams), and a cluster size of 8 children per team. We estimate baseline levels of externalizing problems as measured by the Strength and Difficulties Questionnaire, our main outcome, to be between 50 and 90% in our sample, based on the review of CBT trials for Conduct Disorder or Oppositional Defiant Disorder in Riise et al. (2021). Riise et al. (2021) reports large impacts of CBT among these groups, equivalent to 70% reductions in externalising behavioural outcomes (Gaffney et al., 2021). Few studies reviewed in Riise et al. (2021) focus on 1:1 CBT and the paper argues this average effectiveness could be upwardly biased because of publication bias. The table below shows that we would be able to detect effect sizes considerably smaller than those shown in the literature. Even under a baseline prevalence of externalizing problems of only 50%, we would be able to estimate effect sizes of between 28% and 32% depending on the ICC. If the baseline prevalence would instead be 60%, the MDE reduces to 23% and 25% for ICCs of 0.05 and 0.1 respectively. The final column of the table shows the MDEs under a more conservative estimate of 25 treated and 25 control teams, and the same cluster size. This would allow us to detect effect sizes of between 25% (for a baseline prevalence of 60% and ICC of 0.05) and 36% (for a baseline prevalence of 50% and ICC of 0.1). Prevalence of externalizing problems in both arms at baseline MDE ICC 32 treated teams 25 treated teams 0.05 50% 28% 32% 0.05 60% 23% 25% 0.1 50% 32% 36% 0.1 60% 25% 28%