Subclinical is the right word for that, yeah.
>How is that a complicating factor in discovering causes themselves?
The study links the factor under study to autism diagnosis, rather than some hypothetical "true" rate of autism that includes undiagnosed and subclinical cases. Ten IQ points can be the difference between otherwise-identical autistics barely passing a class versus failing it and getting referred to a psychologist. Or after getting referred, the difference between qualifying for a diagnosis or not.
The latter actually happened to me personally in elementary school: I had held my pen funny in a way that made my hand hurt after a while, and my teacher referred me for refusing to complete a handwritten class assignment. They did a test on various types of school subjects, and I ranked a bit above grade level in some and far above grade level on others. If I was a grade lower in each subject and thus had below grade level scores, I'd've gotten an official diagnosis of some kind of subject-specific learning disability. But instead, there wasn't enough of a problem to need to do stuff, and doing stuff costs money, so they basically shrugged their shoulders and sent me on my way.