- Tension: Standard psychological therapies are assumed to work roughly the same for everyone, but a major study reveals they produce wildly variable and largely disappointing results for autistic adults — with some patients actually getting worse.
- Noise: The default narrative treats therapy non-response as the patient’s problem rather than evidence of a systemic design flaw, while neurodiversity-affirming approaches remain marginal in mainstream mental health services.
- Direct Message: The majority of autistic adults in this study left therapy no better than when they arrived — not because they were treatment-resistant, but because the treatment was never built for how they experience the world.
To learn more about our editorial approach, explore The Direct Message methodology.
A study of autistic adults who received standard psychological therapy for depression and anxiety in England has landed with a finding that should unsettle anyone who assumes the mental health system works roughly the same for everyone: most of the autistic patients showed little to no improvement. The research, published in Nature Mental Health, didn’t just find that outcomes were poor — it found that the system’s standard playbook produces wildly variable results for autistic adults, and that certain populations within that group fared dramatically worse than others.
Clinical researchers have described the pattern plainly: most people’s symptoms stayed largely the same, but some showed gradual or rapid improvement, while others experienced worsening symptoms. That sentence carries more weight than it appears to. It means the therapeutic frameworks that millions of people rely on — cognitive behavioral therapy, guided self-help, counseling — are, for a significant portion of autistic adults, essentially neutral. Or worse.
The scale of the data is what makes this difficult to dismiss. Researchers analyzed records from a large dataset, tracking thousands of autistic adults who accessed routine psychological therapies through England’s mental health services over several years. They followed symptom changes across the first eight therapy sessions using standardized questionnaires — the same instruments used across the broader population. What emerged were distinct clusters of outcomes that tell a far more complicated story than “therapy helps” or “therapy doesn’t.”
For depression, the researchers identified multiple distinct symptom trajectories. The majority of autistic adults in the sample experienced no improvement or only limited improvement. A small group showed rapid gains. Another small group actually got worse. For anxiety, the picture was even more fragmented — multiple different patterns of symptom change emerged, with a small percentage of participants experiencing a worsening from moderate to severe anxiety during treatment. These aren’t edge cases in a tiny sample. This is what standard care looks like for thousands of autistic people moving through the system.
What predicted whether someone would improve or deteriorate? Research suggests that higher difficulty with daily functioning before treatment and struggles with social leisure activities were both associated with worse outcomes. This is where the concept of social camouflaging — the exhausting process by which autistic individuals mask their natural behaviors to fit neurotypical expectations — becomes critically relevant. Camouflaging and the burnout it produces appear to compound the very symptoms therapy is supposed to address. The therapeutic environment itself, designed around neurotypical communication norms, may be asking autistic patients to perform the exact labor that’s making them unwell.
There’s a racial dimension that demands attention, too. According to the study, autistic individuals from ethnically minoritized backgrounds experienced a higher likelihood of worsening anxiety symptoms compared to white participants. That’s not a gap in improvement. That’s an active deterioration — therapy making things measurably worse for people already navigating intersecting layers of marginalization. The finding echoes a broader pattern in mental health research: systems designed with a default user in mind tend to fail people who don’t match that default, and fail them in ways that compound existing harm.
This connects to something I’ve been tracking in health reporting more broadly — the way standardized interventions routinely overlook population-specific realities. We saw it with dietary guidelines that don’t account for how most people actually live, and with financial anxiety frameworks that mistake surface symptoms for root causes. The pattern is consistent: a one-size approach gets treated as universal, and the people it doesn’t fit are treated as outliers rather than evidence that the model needs revision.
Research has shown that autistic adults are significantly more likely to experience mental health conditions like depression and anxiety compared to non-autistic people, and studies have consistently indicated that autistic people generally have lower rates of recovery during psychological therapy compared to non-autistic individuals. What the new study adds is granularity. It’s not just that outcomes are worse on average. It’s that the system produces radically different results for different autistic people, and those differences map onto identifiable factors — functioning difficulties, social demands, ethnicity — that could, in theory, be screened for and addressed.
One of the study’s most practically significant findings concerns timing. Therapy progress could often be predicted by early sessions, according to the researchers. That’s a remarkably early signal. Researchers noted that this may be an important point to review progress and consider adapting the approach if needed. In a system where autistic adults frequently wait months just to access services, the idea that clinicians could identify non-response early and pivot — rather than continuing for eight, twelve, twenty sessions with diminishing returns — represents a concrete, implementable change.
But pivoting to what? That’s the harder question. The researchers emphasize what they call neurodiversity-affirming care — therapeutic approaches that don’t pathologize autistic traits but instead prioritize accessible, culturally responsive interventions. Researchers have argued that the findings underscore the importance of adapting psychological interventions for autistic people. This involves improving accessibility, by adjusting communication and sensory environments. That means rethinking everything from the lighting in a therapy room to the conversational demands of a CBT worksheet to the implicit expectation that a patient will make sustained eye contact while discussing their most painful experiences.
The concept of neurodiversity-affirming care is still fighting for traction in mainstream mental health services. Much of the existing infrastructure — the training pipelines, the treatment protocols, the outcome measures — was built on neurotypical assumptions. Adapting those systems requires more than adding an autism module to a therapist’s continuing education. It requires a fundamental reorientation: treating the autistic experience not as a set of deficits to manage around, but as a different operating system that requires a different interface. The BBC has reported on similar challenges in neurological care, where standard approaches fail to account for neurodivergent populations.
What makes the Nature Mental Health study land differently than earlier work is that it doesn’t just document failure — it maps the terrain of that failure with enough precision to make targeted changes possible. The multiple depression trajectories and anxiety patterns aren’t just statistical artifacts. They’re descriptions of real people sitting in real therapy rooms, some getting better, many staying stuck, some getting actively worse. The early-session prediction window offers clinicians a decision point. The demographic and functional predictors offer screening criteria. The data is there. The question is whether the system will use it.
There’s a quiet brutality to seeking help for depression or anxiety — conditions that already erode your confidence in your own perceptions — and having that help not work. For autistic adults, who may have spent years building the courage to enter a therapeutic space that wasn’t designed for them, the experience of non-improvement isn’t neutral. It reinforces a narrative many have internalized since childhood: that the problem is them, that they’re the ones who don’t fit. This study suggests the opposite. The problem is a system that treats adaptation as the patient’s responsibility — and then measures success using tools calibrated to a population the patient was never part of.
When medicine discovers that a single intervention works across multiple conditions, it makes headlines. When research shows that a standard intervention fails across multiple populations, the response tends to be quieter — a journal article, a conference presentation, a recommendation for “further study.” Thousands of autistic adults moved through England’s mental health services over several years, and the majority came out the other side no better than when they went in. That’s not a finding that needs further study. That’s a finding that needs a different system.
Feature image by SHVETS production on Pexels
