Saturday, 14 May 2011

The South Korean study

It is not that often that a study marks a turning point in autism research. It is something that most researchers in every field of endeavour, whether they admit or not, aspire to produce. Whether out of professional pride or insatiable scientific curiosity or for more personal reasons, the desire for their paper - their hard work - to be hallowed forever after is at the back of the mind.

We have had a few such studies down the years that have perhaps fulfilled this criteria including: Kanner's original 1943 descriptions, the 1979 epidemiological study by Lorna Wing and Judy Gould, the 1995 strongly genetic disorder study, the now-retracted 1998 Lancet bowel problems case series, the 2006 1% prevalence paper, the 2010 AAP position papers on GI dysfunction and autism (here and here) and the recent 2011 1% adult prevalence study. I would perhaps add Asperger's 1944 contribution to this list, if only I could put my GCSE German to good use.

I apologise if I have not included more of your particular choices, but these are the movers and shakers that I remember (not necessarily first-hand I might add - I'm not that old). I would by the way, also say that I am not offering any opinion on whether these studies were right or wrong in their findings, and am also mindful that some people might disagree with the presented choice given the various debates that continue to occur. The point I am however trying to make is that these papers have, in one way or another, secured a place in autism research history. I would also perhaps add a few other studies - not mine - related to my small area of interest in autism research (maybe this one and this one) as a side-line; make of these what you will.

I think we can safely say that the 'South Korean' study will also be added to this list given the quite phenomenal prevalence rate of 1 in 38 children with an autism spectrum disorder reported therein. I say all this mindful of the fact that the figure of 1 in 38 children with autism represents real people and real lives rather than just being some academic exercise. I will attempt to highlight some research points from the South Korean paper in due course but first wish to reflect a little on the human implications of the study now that the media dust is beginning to settle.

There are already some views emerging following the publication of this paper. Views as to what it means to have autism (or rather an autism spectrum condition) and views on what the potential implications are from this study. For some, this study and the astounding prevalence rate, is support for the 'neurodiversity' movement suggesting that autism is a part of the complex tapestry called the human condition. For others this study is support for the growing health issue that is autism, with a vast increase in numbers of cases (people) who will require often significant help and support in the years ahead. The perspectives tend to differ as a result of lots of factors not least a persons' own personal experiences of autism, whether they are a person with autism, a parent or caregiver, professional or policy-maker and what 'autism' they are talking about. Looking at the South Korean study, I would forward the argument that both viewpoints are actually correct; you will see why (hopefully) shortly.

As I posted previously, the study (which will henceforth be referred to as the SK study to save my poor typing fingers), whilst looking at one specific district in South Korea, actually looked at 2 different populations: those in 'regular' school (n=36,592) and those in 'special education' school or on a disability register - a so-called 'high probability group' (n=294). The paper cites a total population of 55,266 7-12 year old children in the abstract (used as the final denominator of prevalence) but about 1 in 4 schools initially approached said they did not want to participate in the study leaving a combined total of 36,886 participants.

A 2-stage design was used during the study; the first stage involved screening children for autistic traits using the Autism Spectrum Screening Questionnaire (ASSQ). For those positive screeners, a subsequent battery of gold-standard, autism-specific assessments (ADOS and ADI) combined with cognitive-intellectual testing was carried out. A similar design was used in the recent 1 in 100 adult study albeit with a different screening instrument.

The results.

  • The initial response rates from the ASSQ screen were quite different (35% high probability vs. 63% general population). Interesting also, that ASSQ data for 314 children were completed by teachers not parents.
  • The crude, unadjusted prevalence rates (unadjusted for non-participants in the study) per group for all autism spectrum disorders were: high probability group = 0.18% and general population = 0.19%. Remember, these were prevalence rates which took no account of those who did not participate in the study and were based on the total population figure denominator of 55,256. (If you calculate prevalence based on those actually sampled, (n=36,866) you get 0.4% and 0.1% per group with a cumulative 0.5% prevalence rate).
  • The adjusted prevalence rates were: high probability group = 0.75% and general population = 1.89%. The cumulative prevalence rate was 2.64%. I have tried to look at how the 'adjustments' were made but fear that my limits have been reached on such statistical challenges.

A few other interesting points are noteworthy.

I made a comment on my first post about this study that because the majority of the prevalence was in the general population group who were in regular school and unrecognised to disability registers, this might imply a large group of children with an autism spectrum disorder with average (or above average) intellectual functioning were driving the prevalence figures. The study text does seem to bear this suggestion out. The percentages of children, for example, in the general population group with autism (core autism) falling into categories of superior and average IQ groups were 11.1% and 55.6% respectively. This is compared with figures of 6.8% and 28.4% in the high probability group.

The flip-side of the coin is the fact that autism, particularly in the high probability group, was accompanied by a considerable degree of intellectual disability; again looking just at core autism, borderline and profound intellectual disability were present in 33.8% and 27.0% respectively (compared with 25.9% and 7.4% in the general population group). The research literature would agree with these figures.

Also, the authors reported finding no cases of Rett syndrome and no cases of Childhood Disintegrative Disorder/Heller's syndrome in the cohort. I am a little bit surprised at this given the large sample size included and the estimated prevalences of such conditions (Rett syndrome figures here); indeed the figures for CDD prevalence come from one of the authors of the SK paper. Does this imply that South Koreans are somehow 'protected' from Rett syndrome and CDD?

So there you have it, 1 in 38 children in this study of autism in a district of South Korea. What this means for other prevalence studies and for the estimates for other parts of the world remains to be seen.
I hope I have made a case for how this study straddles both the neurodiversity and need for more services suggestions and how one way or another, society really does need to sit up and take notice.