Social-Skills Training for ADHD: Weak Evidence at School

What the review found

“We are unable to conclude whether social skills training is beneficial for children with ADHD or whether it is a waste of scarce resources in clinical practice.” That is the closing judgement of a Cochrane review published in 2019 by Ole Jakob Storebø and colleagues at Region Zealand Psychiatry in Denmark, working with the Copenhagen Trial Unit.

They searched eight databases and two trials registers to July 2018 and found 25 randomised trials, reported across 45 papers, covering 2,690 children and adolescents aged five to 17. The programmes ran from five weeks to two years and taught what the name suggests: reading the cues in a conversation, waiting for a turn, recognising what another child’s face is doing. Children in the comparison groups received nothing, or waited.

The review’s primary measure was social skills at the end of treatment, rated by teachers. Across 11 trials and 1,271 children the difference was 0.11 standard deviations, with a confidence interval running from 0.00 to 0.22. Converted to the Social Skills Rating System, the scale most of these trials used, that is 1.22 points out of 102, and a difference has to reach about 10 points before anyone treats it as clinically meaningful.

Teacher-rated emotional competence did not move across two trials, teacher-rated general behaviour did not move across eight, and teacher-rated performance in school did not move across five.

Parents saw more. On parent ratings, social skills improved slightly, general behaviour improved, and ADHD symptoms fell by roughly half a standard deviation. No trial blinded the families or the staff delivering the training, and only two of the 25 used blinded raters, so every parent completing a questionnaire knew whether their child had attended. Storebø and colleagues treated the parent-rated results as the more questionable of the two sets for that reason.

Teacher-rated ADHD symptoms did improve, by 0.26 standard deviations across 14 trials. The finding did not hold up: removing the three longest trials, or the three largest, left no effect, and the trials disagreed with one another substantially.

Cochrane grades how much confidence to place in each estimate, from high down to very low, and here the grade is the substance. Almost every outcome was graded very low, downgraded three levels chiefly for risk of bias and imprecision. Teacher-rated general behaviour reached low. Ten trials measured satisfaction with the treatment, and it was high in all of them; where groups were compared, satisfaction did not differ. Satisfaction is the argument usually made for programmes of this kind, and the review says so.

What it means for you and your child

A treatment shown not to work and a treatment not yet shown to work are different things, and this review sits in the second case. All 25 trials were judged at high risk of bias, more than half were at high risk in how children were assigned to groups, and seven had authors with financial or professional stakes in the outcome. That combination produces an absence where a verdict should be.

One analysis came nearer to a verdict. A trial sequential analysis of teacher-rated social skills, pooling four trials and 185 children, almost reached the boundary past which further trials would be unlikely to reveal a meaningful effect. The reviewers still call for larger and better trials, and note that the evidence on adolescents is weaker again, since only one of the 25 studied them.

Two of the trials looked for adverse events and recorded none. On the question of harm the review is close to silent, and it says as much.

Teachers and parents rated the same children and disagreed. Teachers stand further from the decision to enrol a child, and they observe him in the setting where the trained skills are meant to be used, which is why Storebø and colleagues made the teacher ratings the primary outcome and treated the parent ratings as the weaker set.

None of this settles anything for a particular child. Averages across 25 trials describe populations, and a child who came out of a course steadier did come out steadier. The cost, though, is countable: a standing place in the week, a run of weekend mornings, a fee. What the review cannot supply is the comparison that would justify that outlay, because it examined no alternative against which to set it.

Storebø and colleagues put the phrase “waste of scarce resources” into their own conclusion, which is a severe thing for a Cochrane review to write about a treatment in wide use. Whether a given programme is such a waste, they cannot say. A family whose child likes the group and comes home lighter for it has a reason to continue, and the research is not yet able to supply a second one. Twenty-five trials have been run and the question they were built to answer stands where it stood, which is what a decision made now is being made without.