Addressing Common Problems in Autism Research: Cohort Development and Sample Description for a Participatory, Real-World, Longitudinal Study of Outcomes in Autistic Adults.

A multi-site, community-based participatory cohort study in the US. Three subcohorts were recruited: Healthcare (patients from Oregon Health & Science University and Vanderbilt University Medical Center), Disability Services (Multnomah County, Oregon and San Diego, California), and Community (US-wide via social media, community organisations, and word of mouth). 3,480 people were invited to the healthcare and disability-services subcohorts; 942 consented and 870 completed the baseline survey (319 healthcare, 266 disability services, 285 community). Participants completed 22 outcome measures and 12 participant-characteristic questionnaires at three time points over 12–18 months. A 15% random sample re-took the surveys two weeks after the second time point to check test-retest reliability. Analysis was descriptive: chi-square tests and ANOVA to compare subcohorts and sites. The project was NIH-funded and run as a community-academic partnership with 10 autistic community partners involved in all phases.

This is a cohort description, not an intervention trial, so there is no treatment effect to report. The main result is that the three subcohorts differed significantly in nearly every characteristic. The community group was older, more likely to be women, non-binary, or LGBTQIA+, more independent, and more likely to complete surveys without support (94% vs 50% disability services, 64% healthcare), but had worse overall health, higher rates of depression, anxiety, and PTSD, and lower rates of receiving needed support. The disability-services group was younger, more likely to be male, heterosexual, and cisgender, more often diagnosed in childhood, and had greater support needs. The healthcare group had lower overall health and more co-occurring conditions including mobility disability, Deafness, ADHD, and PTSD. Significant differences also appeared between the two sites within each subcohort. Follow-up retention was 86% at time point 2 and 81% at time point 3. Over 90% of participants who started a survey at any time point completed all applicable sections.

The AASPIRE Measurement Toolkit includes accessible, free measures of adaptive functioning, daily living, and employment that OTs can use to track outcomes for autistic adults. The sample data show that 25% of participants needed support with basic daily activities and 67% needed support with more complex tasks, and that even within disability-services settings some people were not receiving the support they needed. If you are assessing or planning support for an autistic adult, the free caregiver-report versions of these measures may be useful when the person cannot complete a survey directly, even with accommodations.

The toolkit includes accessible communication measures, and the sample data show that 30% of participants found speech difficult or impossible, while the community-recruited group reported greater overall communication fluency yet were the least likely to say they could rely on speech 'all the time.' The authors suggest this may reflect context-dependent barriers (stress, sensory load, social pressure) rather than a fixed deficit, and point to the role of AAC and writing-based communication. Free, accessible communication outcome measures with direct-report and caregiver-report versions are available for SLPs to use in assessment and service evaluation with autistic adults.

Response rates in the systematically recruited subcohorts were low (11%–20%), so the samples may not represent all autistic adults in those settings. Only 73 participants (8%) had a caregiver complete the survey on their behalf, so findings may not generalise well to autistic adults who cannot participate directly even with supports. All data are self-report or caregiver-report; no observational, administrative, or structured diagnostic data were collected. The adapted and original versions of the surveys could not be directly compared to confirm the adaptations improved accessibility or validity. The study is limited to US sites in Oregon, Tennessee, and California, and to participants who could respond in English (with or without supports). This is a single cohort description; the subcohort differences reflect sampling and setting, not causal effects of any intervention.

NIH-funded. Julie Lounds Taylor is an Editor for the journal in which this was published; Christina Nicolaidis is Editor-in-Chief for a different Sage journal. No other conflicts declared.

Do not read the differences between the three subcohorts as evidence that one group of autistic adults is 'better off' or 'worse off' than another. The groups were recruited through different systems (health records, disability-services caseloads, community networks) and the differences largely reflect who those systems reach, not inherent differences between the people. The paper itself warns against overgeneralising from any single sampling frame.