In brief
Autism studies do not reflect all people on the spectrum equally well. In an evaluation of 301 studies, 94 per cent of participants had no intellectual disability, although worldwide about half of people on the autism spectrum are said to be affected. Women, adults, people without spoken language and people outside Western countries also hardly appear in many studies. What studies show therefore does not apply to everyone.
Intellectual disability: a big imbalance
Russell and colleagues evaluated all original studies that appeared in 2016 in autism-specific journals with an impact factor above 3. These were 301 studies with 100,245 participants on the autism spectrum. According to the meta-analysis, 94 % of participants had no intellectual disability (confidence interval 91 to 97 %). It counted as selection bias if more than 75 % of participants had none. This applied to eight in ten studies.
In 38 % of studies the information on intellectual ability was missing entirely. Where there was bias, only 31 % named limited transferability as a limit. The authors call for transparent reporting and strategies to include this group. They point out that the problem can extend to further hard-to-reach groups, such as people with severe intellectual disability, without spoken language or with complex mental co-occurring conditions. What we know about autism might above all reflect the groups that are easier to reach. [4]
Treatment studies: severely affected children are included less often
Stedman and colleagues examined 367 treatment studies with children that appeared between 1991 and 2013. They found that the share of studies that included severely affected children fell markedly over time. Measuring and reporting severity, that is communication ability, thinking and everyday functioning, was very inconsistent. The authors warn of an unbalanced picture that leaves behind those who perhaps have the greatest need, and propose minimum standards for reporting. The analysis concerns only treatment studies in children, and single-case studies with one person were excluded. [5]
High support needs: a perspective of its own
The Lancet Commission on the future of care and clinical research in autism introduced the term “profound autism”. It is not intended as a diagnostic subgroup but as an administrative term to draw the attention of policy, practice and research to the group with very high support needs. In three cohorts, between 18 % and 48 % of people met the criteria. This information comes from a commentary on the commission. The commission itself was not read here. [8]
Further gaps that stand out in the studies on this website
When reading the studies for this project, the same patterns kept standing out. In a meta-analysis of mental therapies, only 7.1 % of participants were women and mostly people without intellectual disability. [1] Studies on AI-supported support had hardly any girls and no adults. [3] Peer support studies come only from Western, English-speaking countries. [7] On loneliness, studies concentrate on younger and middle-aged adults with average or higher abilities. [6] A case study on adaptation at school had mainly white, middle-class girls. [2] These are individual examples and not a systematic evaluation.
What this means for you when reading studies
For every study report you can ask who took part, that is about age, sex, support needs, language and origin. You can ask how the participants were found and whether the study itself says for whom the results apply. These are suggestions from the editorial team. The question of transferability also appears in our news items under “Limits”. Basics on reading studies are explained by the article How good is a study? Evidence explained, on sex and gender the article Autism and sex/gender in research and on participation the article Participatory autism research.
What we know and what is open
What we know. People with intellectual disability are strongly under-represented in autism research, and similar gaps show in treatment studies. [4] [5]
Open.
- How big the gaps are for women, adults, people without spoken language and different origin groups. There is no uniform evaluation as for intellectual disability, and the examples named are individual findings.
- How much the results would change if these groups were included. [4]
- Russell’s study examined only autism journals. Studies in other journals are missing. [4]
- The share of severely affected children in studies fell until 2013. The development afterwards was not examined. [5]
You might also be interested in
- How good is a study? Evidence and recommendations explained
- Autism and sex/gender in research
- Open research questions
- Participatory autism research: research with rather than about people on the spectrum
Sources and literature
- Ding X, Luo H, Zhang J, Yang H, Fan Y, Wu J, Wu S: Comparative effectiveness of non-pharmacological interventions for anxiety, depression, and quality of life in individuals with autism spectrum disorder: A systematic review and network meta-analysis. Frontiers in Psychiatry 2025;16:1660412. https://doi.org/10.3389/fpsyt.2025.1660412 (frei lesbar, CC BY; online 02.12.2025) – Systematic review with network meta-analysis (67 studies, 3,604 participants)
- Goscicki BL, Scoggins ME, Espinosa GH, Hodapp RM: A “Round, Bruising Sort of Pain”: Autistic Girls’ Social Camouflaging in Inclusive High School Settings. Journal of Autism and Developmental Disorders 2025;56(6):2282-2295. https://doi.org/10.1007/s10803-024-06716-5 (frei lesbar, CC BY; online 10.01.2025) – Interview and questionnaire study (31 girls)
- Kuca J, Stencel M, Pilarski B, Florek S, Pudlo R: Artificial intelligence-supported therapeutic interventions for autism spectrum disorder: a systematic review. Frontiers in Psychiatry 2026;17:1833853. https://doi.org/10.3389/fpsyt.2026.1833853 (frei lesbar, CC BY; online 28.08.2026) – Systematic review (14 studies)
- Russell G, Mandy W, Elliott D, White R, Pittwood T, Ford T: Selection bias on intellectual ability in autism research: a cross-sectional review and meta-analysis. Molecular Autism 2019;10(1):9. https://doi.org/10.1186/s13229-019-0260-x (frei lesbar, CC BY) – Cross-sectional analysis and meta-analysis (301 studies, 100,245 participants)
- Stedman A, Taylor B, Erard M, Peura C, Siegel M: Are Children Severely Affected by Autism Spectrum Disorder Underrepresented in Treatment Studies? An Analysis of the Literature. Journal of Autism and Developmental Disorders 2019;49(4):1378-1390. https://doi.org/10.1007/s10803-018-3844-y (frei lesbar, CC BY) – Literature analysis (367 treatment studies)
- Umagami K, Remington A, Lloyd-Evans B, Davies J, Crane L: Loneliness in autistic adults: A systematic review. Autism 2022;26(8):2117-2135. https://doi.org/10.1177/13623613221077721 (frei lesbar, CC BY) – Systematic review (34 articles)
- Verkooijen MH, Ketelaar M, van Woerden M, Staal WG, Tendolkar I, Zinkstok JR: Impact of peer-support programs for individuals with autism: A systematic review. Autism 2025;30(2):300-315. https://doi.org/10.1177/13623613251374971 (frei lesbar, CC BY; online 16.09.2025) – Systematic review (15 articles, 12 programmes)
- Yu X, Xu X: The future of care and clinical research in autism — recommendations from the 2021 Lancet Commission. Medical Review 2022;2(3):216-218. https://doi.org/10.1515/mr-2022-0015 (frei lesbar, CC BY-NC-ND; online 01.06.2022) – Commentary on the Lancet Commission (summary)
You might also be interested in
Knowledge articleScience & research
How good is a study? Evidence and recommendations explained
Knowledge articleScience & research
What counts as success? Measuring the results of support
Knowledge articleScience & research
AI and biomarkers in diagnosis: state of research
Knowledge articleScience & research
Is autism hereditary? Genes, family and risk
Found a mistake?
If something is factually wrong, out of date or unclear, let us know. You do not have to give any personal details.
