What I want to focus on here is a specific type of claim put forth by Blaxill:
Nylander and Gillberg screened adult psychiatric outpatients for evidence of undiagnosed autistic spectrum disorders. This population had not been screened for autism previously. The authors hypothesized that they would find high rates of undiagnosed autism. The screening procedure located 19 adults with autistic spectrum disorders who had not received a prior diagnosis. However, the prevalence in this group was only 2.7 per 10000, a finding that provides little support for a hidden horde hypothesis. The authors note this point reluctantly, claiming that the observed prevalences "should be regarded as an absolute minimum."
Blaxill refers to Nylander & Gillberg (2001). The authors correctly qualify the "absolute minimum prevalence". Every autistic adult would not be expected to be a psychiatric out-patient, as defined in this study. But we can extrapolate. The authors find that 1.44% of these patients had "definite ASD". Now consider that the prevalence of psychiatric disorder in adults varies considerably between 10% and 20%. This would seem to result in a prevalence of 14 to 29 in 10,000, from this population alone.
Note also how Blaxill fails to mention that 89.5% of all autistics found by Nylander & Gillberg were previously "missed".
Scharin & Hellstrom (2004) have found that adult psychiatry does not recognize disorders in the patient population by the frequency the disorders presumably appear.
Consider also Stahlberg et al (2004), which finds that 30% of adult patients with ADHD had comorbid ASD. As of 2006, the prevalence of ADHD in adults in the US is 4.2% [ref]. If Stahlberg's sample is representative, the resulting prevalence of ASD is 126 in 10,000, from this population alone. This is clearly a scandalous result.
Bajerot et al (2001) found that 20% of OCD subjects had "autistic traits" and suggested that OCD is related to HFA and Asperger's. The prevalence of OCD in adults is around 4% [ref]. Putting these numbers together we come up with an ASD prevalence of around 80 in 10,000, from this population alone.
So far we have not even considered the institutionalized population. An early finding by Shah et al (1982) was that 38% of the adult population in a mental handicap hospital had behaviors consistent with autism. A similar result was obtained by La Malfa et al (2004) in an Italian population with intellectual disabilities. Before the study, only 7.8% were recognized as having PDD. Prevalence in the population with intellectual disabilities varies widely from 3% to 50%, according to de Bildt et al (2005). Considering that the prevalence of mental retardation is about 1%, a 38% proportion of ASD in this population would result in an ASD-MR prevalence of 38 in 10,000.
Autism is missed frequently in many populations. For example, among those with bipolar disorder, the prevalence of ASD is anywhere from 8% to 62% [ref]. It is about 32% in the population with epilepsy [ref]. (The prevalence of epilepsy is 0.5% to 1%). ASD affects about two thirds of people with Tourette Disorder [ref]. It is found in anywhere from 10% to 14% of the population with learning disabilities [ref][ref]. (The prevalence of learning disability is around 5% or 6%).
To contrast, note that the current California DDS recognition of autism in the population with epilepsy is about 4%, and 7% in the population with mental retardation. In other words, the prevalence of autism in California has nowhere to go but up.
No studies to date have looked at the total population prevalence of ASD in adults, but one comes close. Baron-Cohen et al (2001) assessed a control group with the Autism-Spectrum Quotient (AQ). It found that 2% scored higher than 32. The study mentions that after an interview, 7 of 11 previously undiagnosed high-scorers met threshold criteria for a DSM-IV diagnosis. It is also known that 20% of autistics are missed by the AQ. Putting these numbers together we come up with a prevalence of 159 in 10,000 for the non-institutionalized population of adults. The 2% finding should be taken with a grain of salt, as the control group size is only 174. But note the study was replicated in Japan by Kurita et al (2005) who found the proportion of high scorers to be 3% in a control group of 215 individuals.
[Errata: The best replication of Baron-Cohen et al (2001) is actually Wakabayashi et al (2004), with a control group of "normal adults" of size 194 and another control group of university students of size 1050. They found that 3% of individuals in both control groups scored more than 33.]
Baron-Cohen et al (2001) also found that scientists (including mathematicians) scored significantly higher than controls, as did the group of Mathematics Olympiad winners. I wonder if these qualify as "geeks who got lucky", which is the other theory Blaxill is fond of.
If these prevalence numbers in adults seem high, consider Posserud et al (2006) who found that 2.7% of children in the 7-9 cohort were high scorers in the ASSQ. This study also documents what I interpret as parent unwillingness to have a child labeled autistic.
So what is the true prevalence of ASD, after all relevant populations are screened? Is it 200 in 10,000 perhaps? Frankly, this question in unanswerable. A diagnosis of autism, to date, is inherently subjective. It's not only the criteria that matters, but also how the evaluator understands and interprets the criteria. Even when more systematic instruments are used, scoring each item is a subjective exercise, and threshold scores are either arbitrary or selected from a prior subjective basis. To take an example, Posserud el al (2006) finds that 2.7% of children score in the 97.3 percentile of the ASSQ. He could very well find that 5% of children are in the 95 percentile. Why is one boundary better than the other?
Timimi (2004) makes a good case about the subjectivity of an autism diagnosis. Other parents might also find some of Timimi's remarks encouraging:
In my clinical practice I often come across children and adolescents who are labeled autistic. When I focus on their abilities I often find much about them that does not fit the autistic discourse. Once I reopen the question of diagnosis many adolescents ask me to officially "undiagnose" them, which their parents are usually very pleased about.
Blaxill should recognize that his "hidden horde" device is of little value at this point. The prevalence of autism in adults is likely about as high as that found in children. Autism is apparently often "missed" in many populations and is only identified after exhaustive screening. Finally, it is important to realize that a diagnosis of autism is not medical, but subjective.
Joseph, father of autistic child