The Relationship of US Autistic Disorder Changepoints to Proposed Environmental and Sociologic Causes

Background: Recently, the EPA (Environmental Protection Agency) published a study that analyzed time trends in the cumulative incidence of autistic disorder (AD) in the US, Denmark, and worldwide.   A birthyear changepoint around 1988 was identified.  It has been argued that the epidemic rise in autism over the past 3 decades is partly due to a combination of sociologic factors – changes in diagnostic rules, improved detection and availability of special education resources.  Without debating whether these proposed sociologic and environmental factors have increased, this study has  calculated changepoints for these factors in order to determine their relevance to autistic disorder. 

Objectives: To compare the relationship between changepoints in autistic disorder (AD) rates and changepoints in proposed sociological and environmental causes of autism.

Methods: US and California AD prevalence or cumulative incidence were collected from previously published articles. Sociologic data were objectively represented as follows: improved detection was represented by the number of professionals who can diagnose autism,   the number of Pubmed publications on autism, and the number of autism-related messages in Yahoo groups; the numbers were normalized to appropriate background counts. Dissemination of DSM revisions was represented by printing dates for each version.  The availability of special education resources was represented by the amount of federal funding per year.  The usage history of a proposed environmental cause, thimerosal in vaccines, was reviewed. When enough data were available, changepoints are defined to be the year when a slope change occurs; piecewise linear regression fits (2 lines or 3 lines) were used to calculate these changepoints.  Otherwise,  changepoints were predicted based on the birthyears first potentially impacted by a specific DSM or vaccine event.

Results: Additional birthyear changepoints were identified from AD data: 1980.9 [95%CI: 1978.6- 1983.1], 1988.4 [95%CI: 1987.8-1989.0] and 1995.6 [95%CI: 1994.6-1996.6], confirming and expanding the EPA work for US data.  AD birthyear changepoints significantly precede the changepoints calculated for indicators of increased social awareness of AD (1997-1999).  Only the 1988 changepoint can be associated with a thimerosal vaccine event, however, low compliance may lessen the importance of that association.

Conclusions: This study confirms the 1988 changepoint detected by EPA and adds 1981 and 1996 as additional changepoints. AD birthyear changepoints, particularly 1981 and 1996, cannot be explained by predicted birthyear changepoints based on altered thimerosal content in vaccines nor on revised editions of the DSM.