Trust Isn’t a Deliverable: Rethinking Ethics in Autism Research That Uses Real-World Data
A recent post from AcademyHealth asks a question many research institutions still avoid: how do you build ethical autism research using real-world data when the systems producing that data have rarely earned our trust?
The post gets more right than most. It opens by challenging the idea that "real-world data" (RWD) is somehow neutral. It’s not. Electronic health records, education metrics, insurance claims — all of it is shaped by institutions that routinely pathologize, mislabel or exclude autistic people. These data systems weren’t designed with us in mind. Often they weren’t even designed to acknowledge we exist.
The authors argue that including autistic people in every stage of research design is both an ethical necessity and a scientific one. Not after the fact. Not as a checkmark. But as co-creators. And they recognize that when autistic people mistrust data collection and analysis, it’s not because we’re resistant to research — it’s because we’ve seen what happens when research speaks about us without us.
Still, the post is cautious. It gestures at co-design but doesn’t name what that requires: shared power, transparent compensation and recognition that trust isn’t a PR strategy. It’s a structural commitment.
Here’s what the blog almost says and what AAB will say outright:
Real-world data is only as ethical as the frameworks interpreting it. A spike in school absenteeism might be read as parental neglect or as system refusal to accommodate. Which lens gets chosen determines who gets blamed and who gets heard.
Autistic people aren’t just data points. We are theorists of our own experience. We know what absence of care looks like. We know what “noncompliance” really means. If you want real insight from real-world data, center us.
You can’t build trust in a system still designed to control us. That means interrogating who gets to define "progress," who names "risk" and whose outcomes matter. That means making room for autistic dissent, not filtering it out.
AcademyHealth has taken a meaningful step in naming the challenge. But naming is just the beginning.
We don’t need invitations to participate in research. We need research designed with the assumption that we were already here.
Because trust isn’t something you build once. It’s something you maintain relationally, rigorously and with humility. Not in abstracts. Not in data sets. But in the structures that decide whether autistic people are seen, heard and believed.