Note: What we’re talking about here = loss of language and other skills between 15 and 30 months of age in children with ASD (Barger et al., 2013).
First, it’s important to understand where data to answer this question comes from. Historically, it’s been from retrospective studies (parent report, home video). More recently, we’ve combined this with prospective studies (e.g. tracking infants at high familial risk of autism from birth, so you can measure as they grow!) The combination of the two is powerful, and has illuminated things not previously captured (e.g. declining visual attention in the first year for kids with ASD (Elsabbagh et al., 2013).
Second, what we’re measuring matters: skills slowly diverging from the typical path is different than a skill plateau, which is different than actual regression—and we have to keep this straight across studies. Also, how you measure it matters—for example, a decline in standard scores on tests over time could represent any of the previously-mentioned phenomena. Further, it’s likely that all three of these things exist in autism—it’s not a homogenous group!
Finally, don’t forget that there are other conditions that are characterized by regression (e.g. Rett syndrome, Heller’s syndrome), and sometimes autism combined with something else (e.g. epilepsy) will show marked regression. So be aware that regression is associated with many childhood disorders.
But, yes, lots of studies point toward there being a notable group of children with autism who show true regression from infancy to toddlerhood. (Do any of you remember how there for a while we thought regression in autism wasn’t a thing, and the parents were wrong? Yeah… The parents weren’t wrong.)
Longer (and obviously the most thorough) answer:
Pearson, N. , Charman, T. , Happé, F. , Bolton, P. F. and McEwen, F. S. (2018). Regression in autism spectrum disorder: Reconciling findings from retrospective and prospective research. Autism Research, 11, 160–1620.
A bit more. We were chatting with Dr. McEwen about this article, and she had a helpful comment we wanted to share:
“…regression seems to occur gradually in some children, and a very slow loss of skills is likely to be harder to pick up than a sudden, dramatic loss of skills. The former might involve a gradual loss of social engagement, whereas the latter could be a child who suddenly loses language. So it's important for speech and language professionals to be aware that it's not just dramatic loss of language that they should be alert to.”