Friday, 28 June 2013

Canada and the autism prevalence rate continued

So, in my last post this morning I was talking about the paper by Hélène Ouelette-Kuntz and colleagues* on the autism numbers game in various regions of Canada. Thanks to Natasa, I now have the full-text of the paper and just wanted to pass some further comments on the findings reported and potentially how important they might be.

  • OK, three regions in Canada were scrutinised based on the NEDSAC initiative for cases of autism: Newfoundland and Labrador (2003-2008 period), Prince Edward Island (2003-2010) and Southeastern Ontario (2003-2010) and a total of 2377 cases of autism were identified across various age groups (2-5 years, 6-9 years, 10-14 years). Response rates - to get more detailed information about the children identified with an autism spectrum disorder (ASD) - per region were not exactly great but ranged between 32% and 45%.
  • The prevalence of autism "increased significantly among all age groups" over the various surveillance periods. There were however some variations across the different regions such that for example, prevalence in the age group 2-5 years old in Southeastern Ontario did show something like a plateau in the later years of study. The authors however do not rule out "incomplete capture of preschool children with autism" as accounting for their results there. Indeed Southeastern Ontario was reported to also have contributed a "larger proportion of false positive" diagnoses based on some small-scale case validation investigation using ADOS or ADI as the standards which may also have accounted for some of the final figure they report.
  • Age of diagnosis and the issue of diagnostic delay are covered in the data presented. Another quote: "in all three regions many children were not diagnosed until they were older". So much for the practical application of all those 'catch 'em early' sentiments.
  • The authors pass comment on the issue of age at diagnosis as affecting any prevalence estimates using the Parner paper** as an example and the suggestion that a younger age at diagnosis would artificially inflate the prevalence rate for that age group. Allowing for potential bias in the reporting of age at diagnosis they concluded: "it does not appear that an earlier age at diagnosis contributed in any meaningful way to prevalence increases in the youngest age group".
  • The sex ratio also came under scrutiny. Interestingly that Icelandic study*** also looking at autism prevalence rates which I mentioned in my last post seemed to take an interest in one of my studies on the topic (see here****). The Icelanders decided that their results did not accord with our own where we talked about an widening gap in the gender ratio towards boys over the years. The Canadian study reported that "the boy:girl prevalence ratio increased between the first and final years of the surveillance period". Indeed in one region Prince Edward Island the ratio went from 3.7:1 in 2003 to 6.7:1 in 2010 (peaking at 6.9:1 in 2009). I'm don't want to say too much about this for now bearing in mind all the possible confounders which could account for results, but certainly if I were an official based somewhere in Prince Edward Island I'd be minded to ask a few questions about the hows and whys of these findings. 
  • The authors admit that the various explanations they've looked at as accounting for the increase in cases "appeared to account for only a small portion of the increases observed". In other words, the numbers are going up but to say that it is better awareness or just that kids are being diagnosed earlier or that it is diagnostic reclassification (which doesn't really cover the period of this study in terms of DSM or ICD changes for example) doesn't really cut the mustard when looking at the size of the increase they uncovered. I'll leave you to determine what else might be able to account for the increase they reported.

I'll finish by reiterating that the autism numbers game is a complicated area. One has to be careful not to make too many sweeping generalisations on how figures, and the potential reasons for those figures, in one area are necessarily representative of autism prevalence in all other areas. 

That being said, the Ouelette-Kuntz paper does perhaps indicate that we should perhaps be widened the debate about why autism is on the rise above and beyond the notion that we're just better at finding it. Yes, there are factors like socio-economic status (SES) to consider but dare I even mention that we might also start to take seriously how environment (however you wish to define this) might also be impacting on the numbers of cases being diagnosed and indeed the concept of gene - environment interactions***** (open-access)?

----------

* Ouellette-Kuntz H. et al. The changing prevalence of autism in three regions of Canada. J Autism Dev Disord. 2013 Jun 16.

** Parner ET. et al. A comparison of autism prevalence trends in Denmark and Western Australia. J Autism Dev Disord. 2011 Dec;41(12):1601-8. doi: 10.1007/s10803-011-1186-0.

*** Saemundsen E. et al. Prevalence of autism spectrum disorders in an Icelandic birth cohort. BMJ Open. 2013 Jun 20;3(6). pii: e002748. doi: 10.1136/bmjopen-2013-002748.

**** Whiteley P. et al. Gender Ratios in Autism, Asperger Syndrome and Autism Spectrum Disorder. Autism Insights. 2010:2 17-24.

***** Chaste P. & Leboyer M. Autism risk factors: genes, environment, and gene-environment interactions. Dialogues Clin Neurosci. 2012 September; 14(3): 281–292.

-----------

ResearchBlogging.org Ouellette-Kuntz H, Coo H, Lam M, Breitenbach MM, Hennessey PE, Jackman PD, Lewis ME, Dewey D, Bernier FP, & Chung AM (2013). The changing prevalence of autism in three regions of Canada. Journal of autism and developmental disorders PMID: 23771514