AuDHD and communication: what holds up, and what does not

Last reviewed August 2026.

I build Subtext, a tool that reads your message before you send it. Which means I spend a strange amount of my week thinking about why sending a message is hard in the first place.

AuDHD kept coming up. Autistic and ADHD in the same person, often pulling in opposite directions. I went looking for research expecting a tidy pile of studies and a list of tips at the end. What I found was patchier and, once I got past the disappointment, a lot more useful than the tips would have been.

So this is what I found. I sell a tool that lives on exactly this problem, so weigh my stake and check the sources for yourself. Everything is sourced at the bottom, and I have flagged the places where the evidence is thin, because a few of the most-repeated claims about AuDHD are not as solid as the internet makes them sound.

First, the uncomfortable part

There is almost no research on AuDHD communication specifically.

That sounds like a strange way to open a research post, but it matters for how you read everything below. Until the DSM-5 came out in 2013, clinicians in most settings were not permitted to diagnose someone with autism and ADHD at the same time¹. The two were treated as mutually exclusive. So the studies that would tell us how the combination works were not being done, and the field is still catching up.

What we do know is that the overlap is large. A meta-analysis pooling 63 studies² put the current prevalence of ADHD among autistic people at 38.5 percent, and the lifetime figure at 40.2 percent. That is a big chunk of autistic people, most of whom were studied as though the ADHD part was not there.

The AuDHD-specific work that does exist is small and qualitative. The one I found most interesting is a 2026 study using in-depth interviews with six women diagnosed in adulthood¹. Six people. You can’t generalise from that, and the author doesn’t claim you can. But the themes are striking, and if you are AuDHD you will probably recognise them. The participants described living in a category the diagnostic system does not really have a name for, and I will come back to what they said about it.

I have to correct something here that gets repeated constantly, including in several of the summaries I read while researching this. You will often read that having both conditions means worse communication difficulties than having either alone. That’s not settled. Salley and colleagues looked at 209 young people using the ADOS³, a clinician-administered assessment, and found the autism-only group scored as more impaired on communication and social interaction than the autism-plus-ADHD group. Two other studies using parent-reported measures found the opposite⁴, and a third found no difference at all. The measures disagree with each other. So does the literature. Anyone telling you the combination is definitively harder is going beyond the evidence.

The finding that reframed the whole thing for me

For most of the history of autism research, communication trouble was treated as a one-sided problem, something missing in the autistic person that needed correcting.

In 2012 Damian Milton proposed something different, which he called the double empathy problem⁵. The idea is that the breakdown runs both ways. Two people with different ways of processing the world misread each other, mutually, and calling that a deficit in one of them is a choice rather than a finding.

Milton’s idea is testable, and it has now been tested. Catherine Crompton and colleagues ran an experiment⁶ where information was passed down a chain of people, telephone-game style. In chains made entirely of autistic people, the information survived just as well as in chains made entirely of non-autistic people. It degraded in the mixed chains. A much larger follow-up, a Registered Report with 311 participants⁷ across Scotland, England and the United States, found that information transfer held up across the board, so the drop in mixed chains from the first study did not show up again. Rapport was a messier picture. Non-autistic chains scored significantly higher than mixed chains, autistic chains barely differed from mixed ones, and rapport rose again when the person disclosed their diagnosis.

A qualitative study published in 2025 pulled the same thread from the inside. Researchers coded 362 excerpts from autistic adults⁸ discussing nonverbal communication on a public forum, and one of the main themes was that the misreading is bilateral, and the autistic people describing it know it goes both ways.

Then there is my favourite study in this whole area, because it is slightly brutal. Sasson and colleagues showed non-autistic observers very short clips⁹ of autistic adults. The observers formed negative impressions within seconds and reported less interest in interacting. When the same content was presented as a transcript, with no audio or video, the bias vanished. The words were fine. The delivery was what got judged. A companion study found that first impressions improved¹⁰ when observers were simply told the person was autistic.

I keep coming back to that pair of results, because they change what “getting better at communication” means. If the content is fine and the channel is what is being punished, then changing the channel is a legitimate strategy, not an avoidance tactic.

The part where you argue with yourself

This is the thing people want explained, and it is the thing the research is worst at. So let me separate what I can support from what I am inferring.

As people describe it, one side of you wants precision, wants to finish the thought properly, wants time before answering. The other side has already interrupted, changed the subject twice, and is now anxious about having interrupted. One participant in that six-person study, Juliette, put it this way:

It very much feels as if there are these two separate parts of my brain - the autistic part and the ADHD part.

The researchers named the theme after her own words: “there are these two separate parts of my brain.” A later, separate theme in the same paper, “two sides of the same coin,” covers close to the opposite idea, that the two conditions can complement each other, and draws on a different line from Juliette about her ADHD “balancing out” her autism.

That is the closest thing to direct evidence I found¹, and it is six people talking about identity rather than about conversation.

What I found better documented sits on the ADHD side, with a catch. Green and colleagues reviewed 30 studies¹¹ and found a consistent pattern of excessive talking, poor conversational turn-taking, and loosely organised speech. A 2021 systematic review connected those patterns to executive function¹², specifically planning, working memory and emotion regulation, and not to language ability as such. Both bodies of work are about children, though, and applying them to adults is a step I am taking, not one the researchers took for me.

On the autistic side, the qualitative work is clear that managing the nonverbal layer of a conversation costs time and energy, and that autistic adults are aware they are spending it⁸. Stack that on top of an attention system that has already jumped ahead and you get the argument.

What I would not do is try to suppress the tangent. There is a related finding I like here. A survey of autistic adults found that special interests were associated with higher subjective wellbeing¹³ and with satisfaction in areas including social contact and leisure, though very high intensity of engagement went the other way. That study is about the interests themselves, not about talking someone’s ear off about them, so I am extending it. But I’d flag the direction, given how much advice treats enthusiasm as a symptom to manage.

Naming the shape of what you are doing seems to work better than trying not to do it. “I’m going to go on about this for two minutes, cut me off whenever.” “I have three things and the second one is the one that matters.” “Sorry, I interrupted, go back.” None of that is in a study. It is just cheaper than the alternative, which is monitoring yourself in real time while also trying to have a conversation.

Channel is the biggest lever you have

If I could only give one recommendation from all of this, it would be to move the important stuff into writing.

The clearest evidence comes from Howard and Sedgewick, who asked 245 autistic adults¹⁴ to rank six communication modes across seven different situations. The paper is called “Anything but the phone!” which tells you most of what you need to know. Phone calls came out worst. Email and text ranked highly, particularly for dealing with services, customer support and other high-stakes low-warmth situations. The authors aimed their conclusion at organisations, not individuals, telling them to stop defaulting to phone calls.

There is a physiological reason underneath the preference. A 2025 study combining behavioural testing and EEG¹⁵ compared 31 autistic and 31 non-autistic adults listening to speech in background noise. The autistic participants showed reduced neural tracking of the speech and a delayed semantic response, even in cases where their accuracy was the same. The comprehension arrived. It just arrived later, and cost more to get there.

Add monotropism to that. Dinah Murray, Mike Lesser and Wendy Lawson proposed in 2005¹⁶ that autistic attention tends to pool deeply into a few channels at a time rather than spreading thinly across many. It’s a theory, not a settled finding, and it is popular in the autistic community partly because it was co-developed by autistic researchers. But it explains the phone problem neatly. A live call asks you to decode tone, monitor your own prosody, hold the thread, manage the pause, and compose a reply, all at once, with no time to think. Text takes almost all of that away. It is also the premise Subtext is built on. Writing gives you a draft, and a draft can be checked before anyone reads it, which no live channel allows.

Practically, that looks like:

• Asking for an agenda before a meeting, and sending a written summary after
• Moving anything with stakes to email or messaging, and saying so plainly (“I’m much better on email, can we do it there”)
• Cameras optional on video calls
• Giving yourself permission to say “let me come back to you on that” instead of answering in real time

None of that is a workaround for a deficiency. In each case you are choosing the channel where your comprehension is not being taxed for no reason.

When you cannot find the word for what you feel

This one took me a while to understand properly, because it’s easy to confuse with not caring.

Alexithymia is difficulty identifying and describing your own emotional states. A meta-analysis of 15 studies found it in roughly 49.9 percent¹⁷ of autistic people, against about 4.9 percent of non-autistic people in the comparison samples. I’d note that the number describes a subgroup, not a universal autistic trait. Roughly half. It also shows up in ADHD, where a case-control study of 101 adults found alexithymia in 41.5 percent¹⁸, with impulsiveness predicting it.

If you are in that subgroup, a lot of standard communication advice becomes impossible. “Just tell them how you feel” assumes a lookup that is not returning anything.

What seems to help is removing the requirement to produce the feeling on demand. Writing instead of speaking, because it gives you the processing time. Picking from a list, so you are not generating from nothing. That’s the gap Subtext is built to fill, by offering a set of versions to choose from instead of a blank page. And treating “I don’t know what I feel about this yet, give me until tomorrow” as a complete and honest message rather than a failure to communicate. Because it is one.

The silence spiral

The pattern I hear about most often goes like this. You send something. They do not reply. Twenty minutes pass. By hour three you have constructed a full account of what you did wrong, long before anyone could know what it means when someone reads your message and doesn’t reply.

The term that gets attached to this is Rejection Sensitive Dysphoria, and I want to be careful here, because RSD is where the AuDHD internet is furthest ahead of the evidence.

RSD is not in the DSM-5. It’s not a formal diagnosis. The widely-quoted claim that 99 percent of people with ADHD experience it comes from the psychiatrist William Dodson¹⁹, and it’s his clinical estimate, not a measured prevalence. As of a 2026 review by a clinical psychologist²⁰, there were about five research studies addressing RSD directly, all qualitative, with samples ranging from 4 to 43 participants. That’s where research on a new construct should start. It’s not where you’d want it to stay before people start describing it as a core feature of their neurology.

What I found much better established is the thing underneath it. A meta-analysis of 13 studies covering 2,535 adults²¹ found emotion dysregulation to be a central feature of adult ADHD, with a large effect size, and emotional lability as the strongest component. The broader literature puts significant emotion regulation difficulty in somewhere between 30 and 70 percent of ADHD adults²⁰. So the experience is real and well-documented. The branded version of it is running ahead of the data.

The practical bit does not depend on which framing you prefer. Ambiguity is the fuel. Two things reduce it:

Set the expectation before you need it. Something like “I read everything, but I sometimes take two or three days to reply properly” does a surprising amount of work. It covers your own reply latency, and it makes other people’s latency feel less like a verdict.

Ask a small, boring question instead of ruminating. “Hey, did my last message land okay?” takes eight seconds and replaces three hours of guessing with actual information. Low stakes, low cost, and it is usually nothing. A question drafted after three hours of ruminating rarely sounds boring, though, and checking what it signals before you send it is the sort of one-line job Subtext is most useful for.

What masking costs

Camouflaging is the effort of performing a communication style that is not yours. Forcing eye contact. Rehearsing small talk. Suppressing the stim, the tangent, the thing you actually wanted to say. Dropping it on purpose is one of the things people mean by autismmaxxing.

The research on what it costs is some of the most serious in this area. A survey of 164 autistic adults found that 72 percent²² scored above the recommended clinical cut-off for suicide risk, and identified camouflaging and unmet support needs as risk markers specific to this group. A mixed-methods systematic review published in 2025²³ reached a similar conclusion, finding camouflaging to be a consistent risk factor for suicidality across studies.

For a long time this was framed as an autism thing. A pre-registered 2024 study compared adults with autism²⁴, adults with ADHD, and a comparison group, and found that the ADHD group camouflaged more than the comparison group, though less than the autistic group. So if you are both, you are likely doing two kinds of masking at once: hiding traits to fit in, and hiding executive slips to avoid getting in trouble. There is no controlled study of what that combination costs, but it’s not hard to guess the direction. It is one reason people give for why a two-line reply can take an hour.

The word for where that leads, in autistic community language, is burnout: chronic exhaustion, loss of skills you previously had, and reduced tolerance for sensory input, typically lasting three months or more. It was defined through participatory research with autistic adults²⁵, not in a lab, and the measures for it are still early. That doesn’t make it less real, only under-studied.

If any of this is describing your own life at the moment, please talk to someone, a friend, a GP, a therapist, anyone. It matters more than anything else on this page.

At work

I found two systematic reviews worth knowing about here.

The first covered 26 studies across seven countries²⁶ with about 7,000 participants and looked at disclosure. The benefits were real, and they included acceptance, accommodations and colleagues who understood what was going on. So were the risks of stigma and discrimination. There’s no universal answer in that data. It depends heavily on where you work.

The second, from 2025, looked at accommodations specifically²⁷ and found them linked to better job stability, satisfaction and productivity, with the caveat that effectiveness depended a lot on relationships and organisational culture rather than on the accommodation itself. Most of the underlying studies were small and self-report, so hold it loosely.

My read of the two together is that disclosure works best when it is attached to something concrete.

Just the label “I’m autistic.”

Hands someone a label and no instructions

Attached to something concrete “I take in written information much better than spoken, so could you send the brief by email and I’ll come back with questions.”

If your workplace culture makes you uneasy about the label, the second one works fine without the first. Getting that second line specific without sounding defensive is the kind of draft Subtext, an AI tone checker for neurodivergent adults, is built to help with.

A tool you will see recommended everywhere is the personal user manual, sometimes called a communication passport: a short document explaining how you process things, what channels work, how you like feedback. I like the idea. I want to be straight that I couldn’t find good evidence it improves outcomes, and a realist review of health passports specifically concluded there is no good evidence they work²⁸, and that they risk becoming a bolt-on that nobody reads. Use one if it helps you. Just don’t assume it’s doing the work on its own.

At the doctor

I’d call this one the exception, because it is the one tool in this whole area with real evidence behind it.

The AASPIRE Healthcare Toolkit and its accommodations report²⁹ were built through community-based participatory research, with autistic adults as co-researchers, and tested with 259 autistic adults and 51 primary care providers. More than 94 percent of patients found it easy to use, important and useful, and early data showed reduced barriers to care and better patient-provider communication.

Beyond that, the practical stuff isn’t complicated. Ask for instructions in writing, say up front that you need a moment to process, bring someone if you can, and mention sensory needs before the appointment.

Three things I would skip

PDA as a settled thing. Pathological Demand Avoidance describes something a lot of people recognise. But it’s not in the DSM-5 or the ICD-11, there is no validation study supporting it as a distinct profile³⁰, and the National Autistic Society notes that the label itself is contested³¹ within the autistic community, with “Persistent Drive for Autonomy” proposed as an alternative. The practical advice attached to it, framing requests as choices and reducing the sense of demand, is reasonable and low-risk. The confident diagnostic framing around it hasn’t been earned yet.

Social skills training as the default fix. The mini-review that covers communication in co-occurring ADHD and autism³² admits that the field mostly recommends assessment tools and training programmes developed for children, then extrapolates. Much of the pragmatic-language research it draws on was conducted with children too¹¹¹². If a programme helps you, good. Just know that the evidence for it in AuDHD adults specifically is close to non-existent.

Any specific statistic about RSD. See above. The experience is real. The numbers aren’t measurements.

A note on the words

I have used identity-first language throughout, so “autistic person” instead of “person with autism.” That reflects the most common preference among autistic adults, but it’s contested. A survey of 3,470 people in the UK³³ found that no single term is universally preferred, and that professionals and autistic adults tend to disagree with each other in a fairly entrenched way.

If you are writing to or about someone specific, the reliable move is to ask them.

If you take one thing

The advice I would give someone, after all of this, is embarrassingly simple.

Move what matters into writing. Say out loud how long you take to reply, before it becomes a thing. Ask the small boring clarifying question instead of running the simulation in your head. And stop treating the effort of performing a communication style that is not yours as a skill you have failed to acquire.

The writing part is where Subtext fits, if you want it. Paste your draft or screenshot the thread, and it names what your message is likely to signal to the person reading it, then offers versions that still sound like you. That last part is deliberate, given what this post says about the cost of performing a style that is not yours. Try Subtext in your browserGet the appTry Subtext in your browser

The research doesn’t say you’re bad at communicating. It says at least part of the difficulty sits between people, and that the channel matters a lot. Those are two very different problems, and only one of them is yours to carry.

---

Autistic, ADHD, or both, and think I’ve read a study wrong? I’d rather know. Find me on LinkedIn.

Samet Durgun is the co-founder of Subtext, an app that catches the emotional tone of your messages and rewrites them in your own voice. He’s based in Berlin.

Sources

Every link above goes to the primary source where one exists.

1. Craddock E (2026). Navigating residual diagnostic categories: The lived experiences of women diagnosed with autism and ADHD in adulthood. Health. Interpretative Phenomenological Analysis, six participants.
2. Rong Y, Yang CJ, Jin Y, Wang Y (2021). Prevalence of attention-deficit/hyperactivity disorder in individuals with autism spectrum disorder: A meta-analysis. Research in Autism Spectrum Disorders, 83, 101759. Meta-analysis of 63 studies.
3. Salley B, Gabrielli J, Smith CM, Braun M (2015). Do communication and social interaction skills differ across youth diagnosed with autism spectrum disorder, attention-deficit/hyperactivity disorder, or dual diagnosis? Research in Autism Spectrum Disorders, 20, 58-66. n = 209, ages 3 to 18, ADOS-based.
4. Summary of the conflicting findings (Rao & Landa 2014, Factor et al. 2017, Harkins et al. 2021 versus Salley et al. 2015) in Unraveling the spectrum: overlap, distinctions, and nuances of ADHD and ASD in children, Frontiers in Psychiatry, 2024.
5. Milton DEM (2012). On the ontological status of autism: the ‘double empathy problem’. Disability & Society, 27(6), 883-887. Theoretical paper.
6. Crompton CJ, Ropar D, Evans-Williams CV, Flynn EG, Fletcher-Watson S (2020). Autistic peer-to-peer information transfer is highly effective. Autism, 24(7), 1704-1712. Experimental diffusion-chain study, n = 72.
7. Crompton CJ, Foster SJ, Wilks CEH, et al. (2025). Information transfer within and between autistic and non-autistic people. Nature Human Behaviour, 9, 1488-1500. Registered Report, N = 311, three countries.
8. Radford H, Reidinger B, Kapp SK, de Marchena A, et al. (2025). “There is just too much going on there”: Nonverbal communication experiences of autistic adults. PLOS ONE. Qualitative analysis of 27 forum threads, 362 coded excerpts.
9. Sasson NJ, Faso DJ, Nugent J, Lovell S, Kennedy DP, Grossman RB (2017). Neurotypical peers are less willing to interact with those with autism based on thin slice judgments. Scientific Reports, 7, 40700.
10. Sasson NJ, Morrison KE (2019). First impressions of adults with autism improve with diagnostic disclosure and increased autism knowledge of peers. Autism, 23(1).
11. Green BC, Johnson KA, Bretherton L (2014). Pragmatic language difficulties in children with hyperactivity and attention problems: an integrated review. International Journal of Language and Communication Disorders, 49(1), 15-29. Review of 30 studies.
12. The profile of pragmatic language impairments in children with ADHD: a systematic review (2021). Development and Psychopathology.
13. Grove R, Hoekstra RA, Wierda M, Begeer S (2018). Special interests and subjective wellbeing in autistic adults. Autism Research, 11(5), 766-775. Survey study.
14. Howard PL, Sedgewick F (2021). ‘Anything but the phone!’: Communication mode preferences in the autism community. Autism, 25(8), 2265-2278. n = 245 autistic adults.
15. Li, Sujawal, Bernotaite, Cunnings and Liu (2025). Auditory and semantic processing of speech-in-noise in autism: a behavioral and EEG study. Autism Research. 31 autistic and 31 non-autistic adults.
16. Murray D, Lesser M, Lawson W (2005). Attention, monotropism and the diagnostic criteria for autism. Autism, 9(2), 139-156. Theoretical paper.
17. Kinnaird E, Stewart C, Tchanturia K (2019). Investigating alexithymia in autism: A systematic review and meta-analysis. European Psychiatry, 55, 80-89. 15 studies.
18. Kiraz and Kartal (2021). The relationship between alexithymia and impulsiveness in adult attention deficit and hyperactivity disorder. Turkish Journal of Psychiatry. 101 adults with ADHD, 100 controls.
19. Dodson W (2019). Rejection Sensitive Dysphoria: Dr. William Dodson brings new insight to Emotional Regulation. Taking Control: The ADHD Podcast, episode 405, interview with Pete Wright and Nikki Kinzer. Dodson states his estimate on the record (“It’s around 99%”), not a published prevalence figure.
20. Rejection Sensitivity Dysphoria: The Actual Research. Psychology Today (2026). A clinical psychologist’s review of the existing RSD literature.
21. Beheshti A, Chavanon ML, Christiansen H (2020). Emotion dysregulation in adults with attention deficit hyperactivity disorder: a meta-analysis. BMC Psychiatry, 20, 120. 13 studies, N = 2,535.
22. Cassidy S, Bradley L, Shaw R, Baron-Cohen S (2018). Risk markers for suicidality in autistic adults. Molecular Autism, 9, 42. Survey of 164 autistic adults.
23. Camouflaging and suicide behavior in adults with autism spectrum condition: A mixed methods systematic review (2025). ScienceDirect.
24. van der Putten WJ, Mol AJJ, Groenman AP, et al. (2024). Is camouflaging unique for autism? A comparison of camouflaging between adults with autism and ADHD. Autism Research, 17(4), 812-823. Pre-registered.
25. Raymaker DM, et al. (2020). “Having all of your internal resources exhausted beyond measure”: defining autistic burnout. Autism in Adulthood, 2(2), 132-143. Community-based participatory research. Accessible overview: National Autistic Society.
26. Lindsay S, et al. (2021). Disclosure and workplace accommodations for people with autism: a systematic review. Disability and Rehabilitation. 26 studies, ~7,000 participants.
27. Heinze (2025). Workplace accommodations and employment outcomes among employees with autism: a systematic review. Cureus. 10 studies, 2010 to 2025.
28. No evidence to show whether autism health passports are effective (2023). The Conversation, reporting a realist review published in PLOS ONE.
29. Nicolaidis C, et al. (2016). The development and evaluation of an online healthcare toolkit for autistic adults and their primary care providers. Journal of General Internal Medicine. 259 autistic adults, 51 providers.
30. Pathological demand avoidance: further research is required (2024). The Lancet Child & Adolescent Health.
31. National Autistic Society, Demand avoidance.
32. Theodoratou M (2024). Communication issues in co-occurring ADHD and autism spectrum disorders. Evaluative approaches and targeted interventions: mini review. Advances in Psychiatry and Neurology, 33(3), 188-195.
33. Kenny L, Hattersley C, Molins B, Buckley C, Povey C, Pellicano E (2016). Which terms should be used to describe autism? Perspectives from the UK autism community. Autism, 20(4), 442-462. Survey of 3,470 people.