A recent paper appearing in PLoS ONE offers some further insight into this process, long-term. The paper (open-text here) by Whitehouse and colleagues* sought to measure autistic traits among typically-developing toddlers and the degree to which such traits were predictive of things a couple of decades later. It is important to realise that this was not a study looking at children with a diagnosis of autism, but rather the overlap of traits in the general population. Longitudinal research to the extreme!
Without repeating the paper rote fashion, here's a summary of what they did and found:
- 2900 mums-to-be were recruited for a study looking at the repeated effects of ultra-sound during gestation. 2868 children were invited to take part in a follow-up study involving various assessments during their childhood and young adult years.
- When children were aged approximately 2 years, their parents completed the pervasive developmental problem sub-scale derived from the Childhood Behaviour Checklist (CBCL) on behalf of their child. The scale looks at social and non-social autistic traits. Participants were followed-up some 17-18 years later and completed the self-report Autism Spectrum Quotient. The scoring schedules whilst not exactly the same, were compared and various potential sources of bias taken into account (most of which occurred during gestation).
- The results from participants (males: 360, females: 400) are presented with correlations between the two instrument scores. A significance value of p<0.006 was used as a result of some statistical adjustment made for the multiple comparisons used.
- Males showed the greatest positive correlations between the testing sessions with several associations surpassing the significance boundaries set in areas of total score and social autistic trait score combinations. Having said that the R-squared values (measure of predictive correlation) were not exactly brilliant (0.16-0.17), where 0 is no predictive value and 1 is absolutely predictive. Non-social autistic traits in males showed nothing in particular across the study. Females scores across social and non-social autistic traits over the study period again showed nothing in particular.
Although there are elements to this study which might raise an eyebrow of many scientific sorts out there, I have to say that I do like this study. I like it because it was prospective and it had a large participant group. I also like the fact after so many years a good percentage of the original group were still willing to volunteer, albeit perhaps from a higher socio-economically advantaged group. Of course we are 'almost' dealing with apples and oranges in terms of the different scales used over the study period and the different informants (parents vs. participants), so these need to be kept in mind.
The gender differences are interesting. Interesting because of the various debates around diagnosis of autism in males and females. The current results hinting that autistic traits in typically-developing young girls were not the same or predictive of the same group in early adulthood may imply either different biological mechanisms or different social factors are at work in modifying such behaviours compared with boys. How this might translate into autism and presentation by gender, I don't know.
The authors are quite reserved in their interpretation of their findings despite the significance values obtained. 'Modest' is a word used quite a bit throughout the article, and to me, that sounds about right. Kudos to them for the work they have put it and the results obtained.
* Whitehouse AJO. et al. Are autistic traits in the general population stable across development? PLoS ONE. August 2011.