“[T]he age at which a person is first diagnosed with autism varies considerably… These differences in developmental timing have conventionally been attributed to differences in severity… Zhang et al. present evidence that the developmental timing of autism diagnosis is not simply an artefact of the challenges of identifying milder cases at early ages, but rather a primary feature that distinguishes distinct forms of autism.”
“Zhang et al. first examined longitudinal data describing socioemotional and behavioural development in three cohorts of autistic individuals. In each cohort, the authors identified two subgroups. These exhibited qualitatively distinct trajectories of development that corresponded with age at diagnosis. A ‘unitary’ model of autism assumes that the developmental timing of diagnosis is a function of severity… Rather, in all three cohorts, difficulty levels in early childhood were similar across the subgroups and increased throughout development only for the later-diagnosed subgroup. By late adolescence, this subgroup presented with more-severe difficulties. This observation is consistent with a ‘developmental’ model that treats timing of diagnosis as a qualitative feature of the condition, as opposed to simply a result of severity.”
“Zhang et al. found that early-diagnosed autism and late-diagnosed autism had distinct polygenic profiles. In other words, specific [genetic] variants associated with risk, and their effects on risk, differed by subgroup.”
“Zhang et al. found that the genetic effects on timing of autism diagnosis persisted at full strength after adjusting for a range of developmental milestones. They also found that distinct genetic factors representing early and late diagnosis contributed independently to autism-case status, even after accounting for genetic factors that underlie a range of psychiatric disorders. The authors further established that their results were not driven by sex differences in genetic profiles of autism risk — a key concern given that girls are often diagnosed with autism later than are boys.”
This is a summary of a recent large-scale analysis by Zhang et al. I find this work rather interesting… as it definitely challenges my assumption that diagnosis timing is just a matter of severity. Also worth mentioning that the late-diagnosed group scored significantly worse on several mental health metrics.
The paper itself (open access): https://www.nature.com/articles/s41586-025-09542-6
In a similar vein, my first thought was “did they control for economic factors?” Lower-income families are less likely to have adequate healthcare, less likely to enroll a child in pre-school or daycare, less likely to see an educator intervene if the child is missing milestones.
Especially in the US, there’s a history of conflating economic factors for genetic factors because of who has historically held wealth. Turns out playing golf doesn’t make you live longer, it’s just that if you can afford to play golf, you can probably afford healthcare and more nutritious foods.
This is an excellent question… they mentioned it in the intro section that “… previous studies have shown that individual clinical and sociodemographic factors explain only a small proportion (typically less than 15%) of the variance in age at autism diagnosis”. They also addressed it a bit deep in the supplementary materials. Copied below:
We ran a few analyses to understand if parental characteristics can impact age at autism diagnosis, primarily through gene-environment correlations.
So I think the answer is yes, they did control for economic factors, and the effect is minimal