The largest synthesis yet, and its two headline numbers
Kamila Castro, Juliana Vaz and colleagues across seven Brazilian federal universities have published, in Autism Research, the most comprehensive quantitative synthesis of quality of life in autism to date: 14,368 records screened, 216 studies included, up to 173 in the meta-analyses, PROSPERO-registered, publicly funded by Brazil’s research council and health ministry. The populations: autistic children and adolescents aged nineteen and under, and their caregivers — with the caregiver analyses alone drawing on more than 28,000 participants.
Two numbers headline it. Autistic young people scored, on the pooled estimate, 21.7 points lower than their peers on quality of life measures standardised to a 0–100 scale (95% CI −26.9 to −16.5). Their caregivers scored 13.64 points lower than comparison groups (95% CI −19.21 to −8.06). Both differences are large, both are robustly significant, and — a finding worth pausing on — the deficit showed no significant difference between children and adolescents. It does not fade with development. Without intervention on the conditions producing it, the gap is stable.
This corpus does not flinch from deficit findings, and it won’t here: the reduction is real, it is measured, and it is substantial. Twenty-two points on a hundred-point scale is not a rounding artefact; it is the difference between a life described as going well and a life described as a struggle, replicated across continents and instruments. Anyone tempted to read the neurodiversity paradigm as a claim that autistic lives are going fine should sit with these numbers. The paradigm’s claim was never that the suffering isn’t real. It is about where the suffering comes from — and on that question, this meta-analysis is quietly eloquent.
Because the finding that matters most is not the size of the gap but its shape.
The shape of the gap — worst where the world makes contact
Break the pooled scores into domains and a pattern emerges that the authors report faithfully and this corpus has been predicting for a year. For autistic young people, the social domain scored worst — a pooled mean of 50.69 — with the emotional domain close behind at 58.9, several studies reporting emotional means below 50. The physical domain was least affected, at 66.0. The caregivers mirror it exactly: social 49.82, emotional 56.3, physical 58.38.
Consider what that ordering means. The physical machinery of these lives — health, mobility, bodily function — is the most intact dimension. What collapses is the emotional and the social: precisely the two domains where an autistic person meets the demands, judgements and sensory environments of a world calibrated to demands no nervous system chose and where a caregiver meets isolation, stigma and an unshared load. A deficit residing wholly inside the person would not distribute this way. A mismatch between a person and their environment distributes exactly this way — heaviest at the interface, lightest in the domains the environment touches least.
The structural moderators point in the same direction. Quality of life rose with national Human Development Index — better-resourced societies, better autistic lives — and the authors are properly careful that this is association, not cause, reflecting healthcare systems, service availability, cultural perceptions of disability and support policies in aggregate. But that is rather the point: if the deficit were fixed in the neurology, it should not vary systematically with the wealth and structure of the society around it. It does. The mechanisms the authors cite from the wider literature — psychiatric comorbidity, barriers to social participation, rejection sensitivity, and the stress of masking and camouflaging autistic traits — are, without exception, factors that live between the person and their context, not sealed inside the diagnosis.
Readers will recognise this as the demand-structure argument arriving in meta-analytic form: the difficulty concentrates where the demands concentrate. The TPAC trial showed a child’s measured severity falling when the surrounding adult’s capacity rose; this synthesis shows, at the scale of 173 studies, that wellbeing tracks the surrounding structure — the domain, the household, the country. Same finding, three magnifications.
The North American paradox — services that reach the child and miss the family
The continental analysis produced the strangest and most instructive result in the paper, and the authors deserve credit for reading it straight.
For autistic young people, North America posted among the higher regional scores — consistent with the most intensive, best-funded, most professionalised child-directed autism service apparatus in the world. For their caregivers, North America posted the lowest quality of life of any continent on earth: a pooled mean of 47.28, against a global caregiver picture in the high fifties and sixties. South America ran the exact inverse — the lowest scores for autistic young people, the highest for caregivers.
The authors’ interpretation: intensive service models may genuinely benefit the child while placing substantial demands on families — the coordination of therapies, appointments, educational advocacy — and direct services aimed at the individual do not necessarily extend to family care or caregiver support. In other words, the most sophisticated autism-services economy in existence has optimised for the child as an isolated treatment target and left the family running the logistics, unsupported, at measurable cost to their lives. The corpus name for this is familiar: it is the REBEL finding — parents as unpaid caregiver-advocate-navigators, doing alone what a village should hold — showing up as the worst caregiver number on the planet, in the system with the most services to navigate.
The lesson is not that services harm families; South America’s high caregiver scores alongside its lowest child scores suggest something like cultural and familial cohesion compensating for structural absence, and neither configuration is the model. The lesson is that both current configurations are half a system — one resources the child and exhausts the family, the other holds the family and under-serves the child — and the meta-analysis makes the case for family-centred planning about as empirically as it can currently be made.
What the instruments can and cannot see
Now the honesty section, because a synthesis this large inherits every weakness of what it synthesises, and the authors list them candidly.
Heterogeneity was extreme — I² around 99% for the major estimates — so the pooled numbers are broad indicators of pattern, not precise population values. Of 216 studies, 184 were cross-sectional: almost nothing here tracks anyone over time, which is why the developmental-stability finding, though consistent, rests on comparing different samples rather than following the same people. Only 48 studies included any comparison group. Publication bias was formally confirmed, with smaller studies reporting larger impairments, meaning the true gaps are probably somewhat smaller than the headline estimates. The four-language restriction skews geography. And the field’s priorities are visible in its architecture: 147 studies focused on caregivers against 80 that assessed autistic young people directly.
That last imbalance points at the deepest issue, which the paper names but cannot solve: much of what is recorded as an autistic child’s quality of life is a proxy report — a parent’s rating of the child’s inner life. The authors cite the known divergence between self- and proxy-reports, and the direction of the problem is documented elsewhere in this literature: proxies, under their own documented strain, rating through their own exhaustion, systematically diverge from what autistic young people say about themselves. A caregiver at 47 points is not a neutral instrument for measuring a child’s wellbeing. The paper’s own closing recommendation is the right one, and it echoes what every instrument-critique in this corpus has concluded: future research must directly involve autistic individuals themselves.
So the fair summary runs: the gap is real, large and stable; its shape — social and emotional, not physical — and its gradients — by development index, by service model — locate its production substantially in the environment; the systems that address it currently reach half the family at best; and the instruments measuring it still mostly ask someone else. A deficit-model reading of these numbers says autism costs 21.7 points of life quality. The evidence-shaped reading says: living autistic, in environments built otherwise, measured by others, costs 21.7 points — and every term in that sentence is a variable.
Citations
Castro, K., Duarte, C. K., Silva, E. et al. (2026) — Quality of Life in Autism: A Systematic Review and Meta-Analysis of Patients and Caregivers — Autism Research
van Heijst, B. F. C. & Geurts, H. M. (2015) — Quality of life in autism across the lifespan: A meta-analysis — Autism
Egilson, S. T., Ólafsdóttir, L. B., Leósdóttir, T. & Saemundsen, E. (2017) — Quality of life of high-functioning children and youth with autism spectrum disorder and typically developing peers: Self- and proxy-reports — Autism
Hayes, S. A. & Watson, S. L. (2013) — The impact of parenting stress: A meta-analysis of studies comparing the experience of parenting stress in parents of children with and without autism spectrum disorder — Journal of Autism and Developmental Disorders
Mason, D., McConachie, H., Garland, D. et al. (2018) — Predictors of quality of life for autistic adults — Autism Research
