(excerpt from Romanczyk, R.G. (1994) Autism. In V.S. Ramachandran (Ed.) The Encyclopedia of Human Behavior. Academic Press: San Diego, Vol. 1, 327-336.)
"The prevalence rate for autism is a controversial topic. Autism has been described as occurring in approximately 15 per 10,000 births. However, some estimates, such as Wing & Gould (1979), place the rate at 2 per 10,000 for the full "Kanner " syndrome, while other estimates are 4.5 per 10,000 (Lotter,1966) or 13.9 per 10,000 by Tanoue et al (1988). Rimland (1971) reports that using his Diagnostic Checklist (Form E-2), a rate of 1 per 20,000 was found for "Kanner" type autism, and that the reliability of diagnosis by independent clinicians was extremely poor. At the other extreme, Wing (1989), in the context of discussing sub-classification of autism, refers to a rate of 27 per 10,000 who display what she terms the "triad of social impairment" (social interaction, social communication, and imaginative behavior).Some current estimates are as high as 20 per 10,000.
The central problem of course lies in the diagnostic criteria. There is a lack of a clear consensus and criteria have changed over time. It is often the case in research on autism that the heterogeneity of the population and the poor diagnostic and subject selection criteria render cross study comparisons most difficult (cf. Kistner and Robbins, 1986). Quite separate from diagnosis and related prevalence are systems used by individual states to identify children in need of specialeducational services. As an example, New York State uses a classification system of 11 "handicapping conditions", one of which is autism. The definition used for classification of autism is "A pupil who manifests a behaviorally defined syndrome which occurs in children of all levels of intelligence. The essential features are typically manifested prior to 30 months of age and include severe disturbances of developmental rates and/or sequences of responses to sensory stimuli, of speech, of language, of cognitive capacities, and of the ability to relate to people, events and objects". As of December 1991 there were 2,613,938 children in public schools and 469,058 in private schools in New York with 2,084 having been classified as having autism (these figures do not include preschool age children). This yields a rate of 6.6 per 10,000 children. Thus depending upon one's point of view and which prevalence estimates one chooses, it could be argued that New York is using a reasonably accurate system of identification, or that children with autism were being significantly over or under estimated."