Student A has the typical motor features of mild cerebral palsy. He walks, but does so awkwardly, and can't climb stairs. At School A, this means he must be carried up the stairs in order to attend his second-floor classes. As a result, he is diagnosed with cerebral palsy.
After awhile, Student A is transferred to School B. His motor skills have not significantly changed, but in School B he can use the elevator to get to his second-floor classes, and he does not have trouble getting around independently. As a result, in School B, he is not considered to have cerebral palsy anymore.
Sounds ridiculous, doesn't it? Obviously, cerebral palsy can't be cured simply by removing environmental barriers. Student A may function better in School B, but he still has cerebral palsy.
But apparently the same is not true of autism. Baron-Cohen et al tested a number of people on a new self-report measure of high-functioning autism, including diagnosed autistics, members of the general public, and university students. He found that a cutoff score of 32 points best distinguished autistic from non-autistic subjects. However, some controls scored above that cutoff. While he couldn't interview the members of the general public (because they were anonymous), he did interview 11 university students with scores above 32. Seven of them met criteria for AS/HFA, but:
"No diagnoses were actually made for two reasons: No parent was present to provide independent developmental data, and because none of those meeting criteria complained of any current unhappiness. Indeed, many of them reported that within a University setting their desire not to be sociable, together with their desire to pursue their narrow or repetitive interests (typically mathematics and computing) was not considered odd, and was even valued. ... In all 11 cases however, there was evidence from self-report of significant impairment in functioning during the school years (social isolation, being bullied, and difficulty in making friendships)."
(I could also complain about requiring parents to give data, as this makes adult diagnosis really difficult sometimes, but that's another issue.) So, basically, they were not impaired enough to be diagnosed, but they had been in childhood*. Note, importantly, that this wasn't because they'd gotten any less weird - they'd just found a setting where weirdness was tolerated.
Is it just me, or is the requirement of 'significant impairment', a criteria for all or almost all conditions in the DSM-IV, really screwy? Hey, we found the cure for autism! Stick them in an environment where being autistic doesn't cause them any problems!
* The four who didn't meet criteria could have been diagnosed with PDD NOS.