There are autistics that clearly have something severely wrong with them. They flap their hand, frequently have extremely low IQs, and are not going to be able to effectively participate in society in any meaningful way. You don't need much any expertise to diagnose these people, other than to know that autism exists. Her expertise came in diagnosing people whom I think many would challenge the diagnosis of. It seemed anybody who exhibited less interest in social interactions or engaged awkwardly, and had 'quirks' was designated as being "on the spectrum."
There are radically different different rates of diagnosis in the US by state. Though there are going to be some biases between states, this does seem at least reasonably strong evidence against any sort of genetic, environmental, or other factor. It's obvious that "autism", whatever it may be be, does exist but I think the sharp differences in diagnostic rates are probably simply down to the clinician choosing to diagnose (or not) people who do not clearly exhibit as autistic. It'd be interesting to have 10 clinicians each independently diagnose the same 1000 children. I expect you'd see some fairly sharp differences between them. Go figure, another aspect of psychology that's not replicable.