How Non-Drug ADHD Treatments Fare Under Blinded Trials

What the statement found

Six trials of neurofeedback, 251 participants, found no significant reduction in ADHD symptoms. Those six were the blinded ones, meaning the people who rated the children afterwards did not know which had received the real procedure and which an imitation. Trials without that safeguard report better numbers.

The figure appears in a consensus statement assembled by 80 authors from 27 countries, led by Stephen Faraone at SUNY Upstate Medical University and published in Neuroscience and Biobehavioral Reviews in 2021. Its 208 conclusions were drawn only from meta-analyses pooling at least five studies or 2,000 participants. Reviews that had not tested for publication bias were excluded, along with treatment reviews whose comparison group was a waiting list or ordinary care.

Those exclusions do most of the work. A child on a waiting list receives nothing, so almost any treatment compared against one looks effective, whereas an active control gives that group something plausible and isolates what the method itself adds. Publication bias is the tendency of disappointing trials to go unpublished.

Cognitive training, in six studies of 287 young people with active controls and probable blinding, produced no significant reduction in ADHD symptoms and no significant effect on maths or reading, though verbal working memory improved moderately across five studies of 263 youths. Social skills training across eleven studies of more than 1,200 young people failed to improve teacher-assessed social skills, behaviour, or grades.

Exercise produced a moderate reduction across ten studies of 300 children that did not survive adjustment for publication bias. Meditation-based therapy produced moderate reductions in six trials of 240 children and adolescents and six trials of 339 adults, but half the trials lacked an active control, and removing the waiting-list studies removed the effect. The authors of that meta-analysis concluded there is “insufficient methodologically sound evidence to support the recommendation of meditation-based therapies.”

Parent training for preschoolers produced moderate reductions in parent-reported ADHD symptoms across fifteen studies of over a thousand participants, and none in the six studies of 403 children where assessment was independent. Cognitive behavioural therapy for adults shows the same split, moderate on self-report across nineteen studies and 896 participants and only small in the two trials with an active control and a blind assessor. The identity of the rater changes the answer as reliably as the control group does.

Medication held its ground. Methylphenidate produced large improvements over placebo on clinician ratings across nine studies of 2,677 young people, and the statement’s summary is that non-medication treatments “are less effective for reducing inattention, hyperactivity, and impulsivity.”

What it means for you and your child

None of this ranks a treatment for one child. Every figure is an average across trials, and most of the research pools samples that are Caucasian or East Asian and weighted heavily towards boys, so the numbers describe populations rather than a particular nine-year-old. Even the treatments with the strongest evidence are only partially effective, and a child who improved on a programme that failed in aggregate did still improve.

What the statement offers a family is a way of reading a claim. Three things determine how much a reported result is worth: what the comparison group received, who rated the child afterwards, and whether the improvement appeared in the trained task or in the child’s ordinary day. Cognitive training illustrates all three at once, since the memory exercises reliably improve performance on memory exercises without moving inattention, maths, or reading. A programme with impressive results against a waiting list and none against an active control has measured attention and expectation, and left its own contribution untested.

A few non-drug findings survive the same scrutiny at modest size. Omega-3 supplementation produced small to medium improvements in ADHD symptoms across three meta-analyses, the largest pooling sixteen studies and 1,408 participants, with no effect on emotional volatility or oppositional behaviour. Restricting synthetic food colours produced a small reduction across five double-blind crossover studies of 164 participants, each child serving as their own control. Organisational skills interventions produced moderate reductions in parent-reported inattention across ten studies of 893 young people, a finding that carries the same reporter problem as parent training.

The statement is a curation of existing evidence rather than new data, and its authors are explicit that they omit well-established findings for which no meta-analysis exists, since “the absence of such a study is not always an indication of knowledge of absence of an effect.” The honest verdict on most of these programmes is that they remain unproven under good conditions, which settles nothing about whether they work. Most of the marketing aimed at parents lives in that gap. How a programme was tested is a fair question to ask, and you are entitled to the answer before you pay for it.