Do the Determinants of Autism Vary By Intellectual Disability? a Critical Review of the Concept of the Autism Spectrum and Its Relevance to Epidemiology

Background:  Individual diagnostic categories of autism have now been subsumed into a single overarching category of an autism spectrum disorder in DSM-V. One of the arguments behind this move was that all types of autism, from Asperger syndrome to childhood autism may have a shared etiology. Although recent large epidemiologic studies support this argument for some putative environmental risk factors, for others, different or even divergent associations have been observed for autism with or without intellectual disability.

Objectives:  The overall objective is to provoke thought and discussion by 1) providing a brief historical overview of the genesis of the concept of the autism spectrum; 2) critically discussing recent findings from a large Swedish cohort differentiating autism by the presence or absence of intellectual disability (ID); 3) reflecting on the utility and the strengths and weaknesses of carrying out epidemiologic studies aimed at studying the ‘autism spectrum’.

Methods: The arguments will be discussed using published and unpublished empirical findings of investigations carried out within the Stockholm Youth Cohort, a large total population register based cohort based in Stockholm County, Sweden. Examples include results of studies investigating parental age, migration, socioeconomic status, parental depression, schizophrenia and bipolar disorder, infections, polycystic ovarian syndrome, gestational weight gain, fetal growth and antidepressant, folic acid and multivitamin use during pregnancy.

Results:  The associations for advancing parental age, poor fetal growth, parental migration, Vitamin D deficiency, and maternal hospitalisation for infections during pregnancy were greater for autism with intellectual disability than for autism without ID. On the other hand, parental schizophrenia, bipolar disorder or maternal depression and antidepressant use have been reported to have stronger associations with autism without ID. Similar estimates of associations for autism with and without ID have been reported for parental socioeconomic status and gestational weight gain.

Conclusions:

The advantages of studying autism spectrum disorders as a combined group including larger samples and increased statistical power to study relationships with rare exposures will be discussed. On the other hand, it is possible that findings may be diluted where the associations differ markedly by the presence or absence of intellectual disability. The implications for the epidemiology of autism will be discussed.