Guide 02 · Why the evidence is thin
Why does so much autism therapy seem to have “not enough evidence”?
A plain-language look at the state of autism research — and why it is so hard to do.
What this is
When you look up autism therapies, the honest answer is often “we're not sure yet.” This page explains two things: (1) why so many therapies land on “not enough good evidence,” and (2) why studying autism is genuinely one of the harder jobs in medical research. Both have been written about directly by scientists.
Question 1 — why the evidence is thin
Large teams have checked the whole body of autism-therapy studies at once. Their conclusion is consistent: many therapies show small hints of benefit, but the studies are usually too weak to be sure. The most careful review of early-childhood therapies found that when you keep only the strongest studies and the most trustworthy ways of measuring change, most of the apparent benefits shrink or disappear.
A few plain reasons this keeps happening:
- The studies are usually small, and many have no comparison group getting a fair alternative.
- Very often the people rating whether a child improved are the parents or teachers who know which therapy the child received — understandable, but it can tip results toward looking better than they are.
- Follow-up is short, so we rarely know if gains last.
- Harms and side effects are hardly ever tracked, so we can't weigh good against bad.
- In a large share of early-therapy studies, the people who created or sell the therapy were involved, and those business ties were usually not fully disclosed.
None of this means the therapies are fake or that families are foolish. It means the proof is not yet solid — a very different message from “this works” or “this fails.”
Question 2 — why autism is so hard to study
Autism is not one thing.
“If you've met one autistic person, you've met one autistic person” is a real research problem: a therapy that helps one child may do nothing for another, so studies that lump everyone together wash out.
No agreement on what to measure.
One review of autism trials found hundreds of different outcome measures, most used only once. When everyone measures something different, results can't be pooled or compared.
No agreement on the goal.
Many autistic adults argue the goal should be wellbeing and support, not making autistic people act non-autistic. That disagreement is about values, which science alone can't settle.
Funding doesn't match family priorities.
Families say they want research on everyday life, services, and mental health; a large share of funding has gone instead to biology and causes.
The people who need the most support are studied the least.
Autistic people who don't use much speech, or who also have an intellectual disability, are often left out — so even good evidence may not apply to them.
How sure are we about all this?
Fairly sure. This is not one lonely opinion — several independent teams, using the standard tools for grading evidence, reached the same picture: the autism-therapy evidence base is mostly low-certainty, and the reasons are well understood. We can be reasonably confident the honest status of many therapies is “uncertain” — and that this is a statement about the studies, not about your child.
What this does and does not mean
Be a little cautious of anyone promising guaranteed results, and ask what kind of evidence a claim rests on.
Stop therapy, or that nothing helps. Some approaches show real, if modest, promise, and the right support can matter a great deal for a specific child.
The gaps are the field's fault, not yours. You are navigating a system that has not yet done the studies you deserve.
This is why you'll see so many therapies in our research library marked Limited — a label that covers both “well-studied and not shown to help” and “too little study to know.”
Honest limits of this page
It draws mostly on English-language research from the US, UK, and Australia — itself one of the field's blind spots. It summarizes the overall picture; it does not tell you whether a specific therapy is right for your specific child.
The choice stays with your family.
Knowing the evidence is thin is not a reason to give up — it is a reason to ask good questions and to weigh what you see in your own child alongside what the studies can and cannot promise. You know your child best, and the choice is yours.
Where this comes from
The main reviews and reports behind this page. Several independent teams reached the same picture.
- Sandbank M, et al. Project AIM: updated systematic review. BMJ. 2023;383:e076733. PubMed
- Sandbank M, et al. Project AIM. Psychological Bulletin. 2020;146(1):1–29. PubMed
- Trembath D, et al. Non-pharmacological interventions: umbrella review. Autism. 2023;27(2):275–295. Journal
- Fusar-Poli L, et al. Umbrella review of psychosocial interventions. Molecular Psychiatry. 2022.
- French L, Kennedy EMM. Annual Research Review: early intervention RCTs. Journal of Child Psychology and Psychiatry. 2018;59:444–456.
- Bottema-Beutel K, et al. Adverse event reporting. Autism. 2021;25(2):322–335.
- Bottema-Beutel K, et al. Conflicts of interest. Journal of Child Psychology and Psychiatry. 2021;62(1):5–15. PubMed
- Provenzani U, et al. Outcome measures in ASD trials. Autism. 2020;24:274–284.
- Lord C, et al. Lancet Commission on autism. The Lancet. 2022;399:271–334.
- Pellicano E, den Houting J. Annual Research Review: shifting to neurodiversity. Journal of Child Psychology and Psychiatry. 2022;63:381–396.
- Pellicano E, Dinsmore A, Charman T. What should autism research focus upon? Autism. 2014;18:756–770.
- James Lind Alliance Autism Priority Setting Partnership (Autistica). 2016.
- Russell G, et al. Selection bias on intellectual ability. Molecular Autism. 2019;10:9.