The number, and the honest uncertainty around it
Lochie Timms-McLean and colleagues at Federation University and the University of Melbourne published a systematic review in Addictive Behaviors in June 2026 estimating the prevalence of ADHD and autism among people experiencing gambling harm. Pooling 14 observational studies covering 3,888 observations, their meta-analysis returned a prevalence of ADHD in people experiencing gambling harm of 24% — against a general-population adult ADHD prevalence of around 2.58%. On the face of it, that is close to a ninefold concentration: ADHD appearing roughly nine times more often among people harmed by gambling than in the population at large.
The figure is striking, and it would be easy to lead with the ninefold multiple and stop there. But the review itself is careful, and honesty about the evidence requires carrying that care across rather than discarding it for a sharper headline. The heterogeneity across the included studies was very high — an I² of 93.7%, meaning the individual study estimates varied far more than chance would predict. The authors ran a funnel plot and an Egger’s test and found significant evidence of publication bias. They state plainly that the pooled 24% should be treated as an approximate summary rather than a population prevalence, and that the true figure is likely to vary considerably across settings and may be lower than the pooled estimate suggests.
So the honest claim is narrower than the headline number, and stronger for being honest: ADHD is substantially over-represented among people experiencing gambling harm — probably several-fold, on a consistent direction of effect across studies — but the precise multiple is uncertain, and anyone quoting “nine times” as a settled fact is overreading a deliberately cautious review. What is not uncertain is the direction. Every analytical cut the authors ran pointed the same way: more ADHD among the gambling-harmed than in the general population. The magnitude is contested; the relationship is not. That is enough to build on, and it does not need inflating.
The autism picture is thinner still. Only 2 of the 14 studies reported autism data — too few to meta-analyse. One found suspected autism in roughly 29% of 96 participants experiencing gambling harm; the other found only 9 of 359 screened positive. The authors are explicit that the autism evidence base is too small to support strong claims, while noting that what little exists points toward elevated risk. The result is a review that can say something firm about ADHD, something tentative about autism, and something important about both: this is a population the gambling-harm literature has barely begun to count.
Why the susceptibility is structural, not a failing
The mechanisms the review proposes for the ADHD–gambling link are worth stating precisely, because the wrong framing turns them into a story about weak willpower and the right framing turns them into a story about design. The authors point to differences in executive function and impulsivity, and to dopaminergic reward processing — the same reward-system characteristics that run through the ADHD literature more broadly.
Take the executive-function point first, because it is the one most easily moralised. The review notes that differences in working memory and planning may make it harder to maintain and implement pre-commitment strategies — setting a limit, sticking to a stop plan, avoiding high-risk contexts — particularly under emotional arousal. Read carelessly, that sounds like a deficit in self-control. Read accurately, it is something else: the protective behaviours that gambling harm-reduction depends on are precisely the behaviours that require sustained executive monitoring and follow-through, and they are therefore precisely the behaviours an ADHD profile finds most effortful. The harm-reduction model assumes a particular kind of executive capacity as the baseline. People who diverge from that baseline are not failing the model. The model was built without them in mind.
This is the demand-structure argument in one of its clearest forms. The ADHD substrate is not defective at resisting gambling — it is a substrate whose reward and executive systems work in a particular way, meeting an activity engineered to exploit exactly those systems. Gambling products strongly engage reward-related dopaminergic pathways; an ADHD reward system is differently calibrated to those pathways; the result is heightened reinforcement and persistence. None of that is a moral weakness in the gambler. It is a predictable interaction between a specific neurology and a specific product, and the product was designed to produce that interaction in everyone — it simply produces it most powerfully in the people whose reward systems it most effectively engages. The susceptibility is not located in the person. It is located in the meeting between the person and a thing built to capture them.
The comorbidity data underline how high the stakes are. Among the studies reporting it, substance dependence ran at 54% in one ADHD-gambling sample and 46% in another. The review also flags the elevated suicide risk that ADHD, autism, and gambling harm each carry independently, compounded where they co-occur. This is not a population experiencing an inconvenient habit. It is a population at the sharp end of multiple overlapping harms, concentrated together, and largely unrecognised as a group.
The industry is designed around the vulnerability
The review’s abstract contains a phrase that does a great deal of work: it raises concern about the “targeted predatory gambling practices of operators.” That is the authors naming, carefully, the thing the prevalence data points toward. If a particular neurology is more susceptible to gambling harm because gambling products engage its reward systems more effectively, then an industry optimising its products for engagement is — whether by intent or simply by the logic of optimisation — optimising them for the capture of that neurology. The vulnerability and the design are not two separate facts. They are the same fact seen from two ends.
The review situates this within the emerging commercial-determinants framing of gambling harm, citing Rintoul and colleagues’ 2025 work under the title “Playing to extinction” — a framing that locates the cause of gambling harm not in the failings of individual gamblers but in the commercial systems engineered to produce and extract from harm. Global gambler losses are projected to reach around 700 billion US dollars by 2028. An industry of that scale, increasingly online, refining its products through continuous behavioural optimisation, does not need a deliberate strategy to target neurodivergent people for the effect to occur. It needs only to optimise for engagement and persistence, because the people whose reward and executive systems make them most engageable and least able to deploy pre-commitment strategies will be disproportionately caught by definition.
The structural reading turns the standard story inside out. The conventional frame treats the neurodivergent gambler as a vulnerable individual who needs better self-control, more education, and stronger personal limits — interventions aimed at the substrate. The commercial-determinants frame, which the review’s own data supports, treats the harm as manufactured at the point where an exploitative product meets a susceptible neurology, and locates the intervention where the harm is actually produced: in the product, the advertising, the design. My corpus has argued this across domains — that what gets framed as individual deficit is the substrate meeting a structure built against it. Gambling is the case where the structure is not merely indifferent but commercially motivated to maximise exactly the engagement that harms this population most. The round hole is not passive here. It is a business model.
The prevention system doesn't count them
The review’s central practical finding is the one that should change something, and it is an absence rather than a presence. Current gambling policy, research, and treatment priorities, the authors conclude, do not consider ADHD and autism as priority populations. The group most concentrated in gambling harm — the group for whom the standard harm-reduction strategies are least suited, because those strategies assume the very executive capacities this group diverges from — is not recognised as a distinct population in the systems built to prevent gambling harm.
This is the apparatus failing in the specific way it tends to fail. It is not that the prevention system actively excludes neurodivergent people; it is that it was built around an assumed default gambler, with default executive function and default reward processing, and it never asked whether the people most harmed might differ from that default in ways that make its interventions miss them. The pre-commitment tools, the self-exclusion schemes, the responsible-gambling messaging — all of them lean on sustained self-monitoring and follow-through, and all of them therefore work least well for the population most affected. The system’s design and the population’s need are mismatched at exactly the point of greatest harm, and the mismatch has gone unexamined because the population was never counted.
The authors are careful, again, about what their evidence licenses, and so should anyone reporting it be: the prevalence estimate is approximate, the autism data are thin, causation is not established, and more research is needed before firm population figures can be stated. But the absence of these groups from prevention priorities does not require a precise prevalence figure to be a problem. It requires only the consistent, multiply-replicated direction of effect the review establishes — that ADHD and autistic people are over-represented among the gambling-harmed — and the structural observation that the prevention tools were built for someone else. You do not need to know the exact multiple to know that a harm-reduction system assuming executive capacities this population diverges from will systematically under-serve them. The number is uncertain. The mismatch is not.
What the review ultimately documents is a population sitting at the intersection of an exploitable neurology, an industry optimised to exploit it, and a prevention apparatus that does not recognise it exists. The honest version of the finding is not a frightening multiple but a structural diagnosis: gambling harm concentrates in ADHD and autistic people not because of anything wrong with them, but because the product is built to engage exactly their reward systems, the harm-reduction tools are built around executive capacities they diverge from, and the prevention system has not yet counted them as a group worth protecting. The substrate is doing nothing wrong. Everything built around it is.
Citations
Timms-McLean, L., Nguyen, H., Thomas, S. et al. (2026) — A systematic review of the prevalence of ADHD and autism among those experiencing gambling harm — Addictive Behaviors
Rintoul, A., McLaren, S., Shandley, K. & Klein, B. (2025) — ‘Playing to extinction’: The commercial determinants of gambling-related harm, suicidality and suicide — The Lancet Regional Health – Western Pacific
Wardle, H., Degenhardt, L., Marionneau, V. et al. (2024) — The Lancet Public Health Commission on gambling — The Lancet Public Health
Song, P., Zha, M., Yang, Q. et al. (2021) — The prevalence of adult attention-deficit hyperactivity disorder: A global systematic review and meta-analysis — Journal of Global Health
