Alan Zarembo is right that there is substantial variability in rates of diagnoses, and that we can't confuse diagnoses with the actual incidence of disease.
Nevertheless, impressions are not the same as a scientific analysis. I published the first quantitative analysis of how much of the increase in diagnosed cases in California could be explained by artifacts (changes in criteria by DSM; earlier ages at diagnosis; inclusion of milder cases) using the California data, which has a long enough history to provide data over many years, and the numbers simply do not add up (Hertz-Picciotto & Delwiche 2009). Combining my results with those of Peter Bearman, it appears that about half of the increase in diagnoses in California are due to changes in diagnostic criteria or practices, but that still leaves about a 3-fold increase that is not explained, and that was as of 2007. Diagnoses in California have continued to rise both in areas with low rates and in areas with high rates. Zarembo is interested in explaining the geographic variation, but the explanations for variation spatially are not necessarily the explanations for variation over time.
This quote is particularly misleading:
No study points to an environmental reason for the worldwide explosion in cases over the last two decades.Given the slow pace of genetic change in large populations, genes can't account for the surge either.
That suggests the explanation for the boom lies mainly in social and cultural forces, notably a broader concept of autism and greater vigilance in looking for it.
First, the wording is telling: 'an' environmental reason. There will not be a single reason for the increase, just as there is no single cause of autism. It is far too complex. But the logic that leads to the third sentence escapes me. How many studies have been done of environmental causes? Very few!!! And of these, most were extremely poor studies - very small samples, or lacking individual-level data (just trying to draw correlations - a study design that is generally not even published in the top epidemiology journals). Is it surprising we've uncovered few leads? The funding for environmental factors has been paltry - a mere $40-50 million in the last 10 years, while more than $1 billion has been spent on genetics. So to conclude that we should abandon looking at environmental factors and turn to social/cultural factors is nonsensical.
Moreover, to the extent the increase is due to diagnostic differences, we need to find explanations for both the increasing numbers of diagnoses and for the autism that has been around "all along." In fact, data are emerging about quite a number of environmental factors. Just this year, major papers were published suggesting a role for traffic-related air pollution (Volk et al 2011), for maternal periconceptional nutrition (Schmidt et al 2011), and for maternal metabolic conditions (Krakowiak et al, in press). Earlier papers have indicated associations with pesticides (Roberts et al 2008; Eskenazi et al 2008).
With regard to social factors, most of the cases in our autism study (The CHARGE Study, which is one of the largest to date) are low functioning, and to have been eligible for California services, they had to have three functional disabilities, meaning they are not the high functioning types of ASD or Asperger's that are often described in discussions about the broadening of what we are calling autism.
With regard to genetics: the largest study to date looking at twins shows that 38% of concordance is due to genetics and 58% to shared environmental factors. This completely overturns the widespread assumption that has driven research for the last decade or more, which was that autism heritability was over 70%. (The paragraph in one of the series' articles about twin studies is out-of-date).That mistake is most curious, since Zarembo earlier wrote about the more recent twin study