al., 1999; California Department of Developmental Services, 2000). Studies reporting much higher rates were from relatively small samples or from state surveys, in which an educational label of autism was associated with provision of intensive services and thus highlight the need for further, well-designed investigations. For example, the Department of Education’s Office of Special Education Programs (OSEP) could support a research study examining the prevalence and incidence of autism, using OSEP data gathered for school-age children since the autism category was recognized in 1991. This study could investigate in particular whether the dramatic increases in numbers of children served with autistic spectrum disorders are offset by commensurate decreases in categories in which children with autism might have previously been misclassified.
The symptoms of autism are often measurable by 18 months of age (Charman et al., 1997; Cox et al., 1999; Lord, 1995; Stone et al., 1999; Baird et al., 2000). The main characteristics that differentiate autism from other developmental disorders in the 20-month to 36-month age range involve behavioral deficits in eye contact, orienting to one’s name, joint attention behaviors (e.g., pointing, showing), pretend play, imitation, nonverbal communication, and language development (Charman et al., 1997; Cox et al., 1999; Lord, 1995; Stone et al., 1999). There are three published screening instruments in the field that focus on children with autism: the Checklist for Autism in Toddlers (Baird et al., 2000), the Autism Screening Questionnaire (Berument et al., 1999), and the Screening Test for Autism in Two Year Olds (Stone et al., 2000).
The Checklist for Autism in Toddlers (CHAT) (Baron-Cohen et al., 1992; 1996) is designed to screen for autism only at 18 months of age. From both the initial study of siblings of children with autistic disorder and from a larger epidemiological study involving a population study of 16,000 18-month-old infants (excluding children with suspected developmental delays), virtually all the children failing the five item criterion on the CHAT administered twice (one month apart, the second time by an experienced clinician and including other sources of information) were found to have autistic disorder when diagnosed at 20 and 42 months (Baron-Cohen et al., 1992; Baron-Cohen et al., 1996; Charman et al., 1998; Cox et al., 1999). However, the CHAT was less sensitive to milder symptoms of autism; children later diagnosed with other autistic spectrum disorders did not routinely fail the CHAT at 18 months. Follow-up of the cases at age 7 revealed that this instrument had a high specificity (98%) but relatively low sensitivity (38%) (Baird et al., 2000), suggesting that it is not appropriate for screening.
The Autism Screening Questionnaire is a new 40-item screening scale