Epidemiology of autism: contrasting differences

One of the more animated debates in autism research relates to the possible reasons behind the quite staggering increase in cases being diagnosed world-wide over the past 25 years or so. One of the benchmark studies used for prevalence of autism (note prevalence not incidence), at least in the UK, is the study published by Gillie Baird and colleagues back in 2006 suggesting a total prevalence of all autism spectrum conditions of 116.1 per 10,000.

Two recent reports caught my eye. The first was a comparison of autism prevalence rates in Denmark and Western Australia published by Parner and colleagues. As per the link they reported differences between Danish and Australian prevalence rates of total autism spectrum conditions (68 vs. 51 per 10,000 respectively). These rates are different. Not only between the two countries in question, but also compared with the Baird data. One could perhaps suggest that there were differences in the birth years covered or methodologies employed to "find" and "record" cases (whether it was actively searching for cases of merely reliant on database figures as much of the Danish data tends to be for example because of the way their healthcare records are set up). This could perhaps explain some difference; my question however would be how much can it explain? Does this mean that prevalence of autism spectrum conditions is higher in the UK, than in Denmark or Western Australia? If so, what are the factors influencing such a difference?

A second study that deserves some comment is a follow-up study from Barnevik-Olsson and colleagues on the prevalence of autism in Somali children living in Sweden born between 1999-2003. They reported prevalence of autism as a percentage comparing Somali children with non-Somali children (rates were 0.98% vs. 0.21% respectively). Their initial study published in 2008 showed rates of: 0.7% vs. 0.19% (Somali vs. non-Somali) for children born between 1988-1998. These data again show differences. Differences in the rates between Somali and non-Somali children with autism, differences between the birth years and also differences with the other studies listed in this post. The issue of whether autism rates are elevated in Somali children is the focus of some interest already so I am not going to speculate on any reasons why there may be a difference.

Going back to the Baird data, again with all the caveats regarding case ascertainment et al, are we actually saying that the prevalence of cases of autism spectrum conditions in Sweden (amongst non-Somali children) are so markedly different from the UK data? Questions, questions, questions.