Parent-Reported Prevalence of Autism Spectrum Disorders in US-Born Children: An Assessment of Changes within Birth Cohorts from the 2003 to the 2007 National Survey of Children's Health

Background: The reported prevalence of autism spectrum disorder (ASD) from the 2007 National Survey of Children’s Health (NSCH) (11.0/1000 children) was twice the prevalence of autism from the 2003 NSCH (5.5/1000). Similar trends were observed in a separate US population-based ASD surveillance system. Three general explanations for the observed increases should be considered: survey measurement changes between time points (internal measurement effects);  changes in ASD awareness, screening, access to diagnostic services, diagnostic criteria, and special education placements that led to increased  identification of children with ASDs (external population identification effects);  and changes in ASD genetic susceptibility among persons having children within a population and/or non-genetic risk factors (etiologic genetic or environmental effects). Because the NSCH covers a wide age range, prevalence estimates from 2003 and 2007 can be compared within subgroups based on either child’s age or birth year.  Comparison of population prevalence within birth cohorts is particularly informative as it holds genetic and non-genetic prenatal and early postnatal etiologic exposures constant and thus allows for assessment of measurement and identification impacts.

Objectives: From the two independent NSCH samples, we examined ASD prevalence in children who were US-born between 1990-2000.  We assessed prevalence differences across surveys both by child age at survey and within one-year birth cohorts.

Methods:  In the 2003 NSCH, prevalent cases were based on parental report of an “autism” diagnosis sometime during the child’s life.  In the 2007 NSCH, question verbiage was revised and expanded.  Thus, prevalence was defined two ways: parental report of an “ASD” during the child’s life and parental report of ASD currently.  Estimates were weighted to be nationally representative of US-born children in various years.

Results:
In 2003, children aged 7-8 years had the highest “autism” estimates (10/1,000 children).  Much lower estimates were observed for children aged 3-6 years (prevalence range 3-5/1000) and 11-13 years (prevalence range 4-6/1000).  In 2007, children aged 5-13 years all had notably higher age-specific prevalence estimates of both “ever ASD” (prevalence range 18-29/1000) and “current ASD” (range 12-17/1000) than the “ever autism” estimates for children of comparable ages in 2003.

For children in each of the 1997-2000 birth cohorts (which includes children ages 3-6 years in 2003), relative differences between 2003 and 2007 estimates were 300%- 600% when comparison was based on the 2007 “ever ASD” definition and 200%- 400% when based on the 2007 “current ASD” definition.  For children in the 1990-1996 birth cohorts (which includes children ages 7-13 years in 2003), the relative differences between 2003 and 2007 estimates were lower but still >100% for 5 of 6 cohorts when comparison was based on the 2007 “ever ASD” definition and >80% for 3 cohorts when based on the 2007 “current ASD” definition.  (p<0.10 for all of these differences and p<0.05 for most)

Conclusions: Prevalence differences within birth cohorts are likely partially attributable to survey question changes. However, the magnitude of most differences suggests that continuing identification of children with ASDs between surveys was sizable, even among the oldest children, 7 years or older in 2003.