Detecting autism early?

Early, accurate detection is one of the holy grails of autism research; whether through some operationalised behavioural criteria or the development of an objective biological test (like the Guthrie test for PKU). I say all this because of new piece of research has been published carrying a few headlines like this from the BBC: Screen may spot autism in one-year olds. The paper abstract can be found here and the checklist used - Communication and Symbolic Behavior Scales Developmental Profile - can be download here.

A similar paper has also appeared in JADD on the development of a brief screening tool: The Autism Mental Status Examination. It is not exactly the same thing but perhaps seems to complement the other screening tools such as CHAT and M-CHAT which have been put forward for a screening role at 18 months of age. The article by Pierce and colleagues describing the CSBS-DP is interesting. Interesting insofar as the initial yield of the screening instrument based on 10,479 infants at a 12 months screen, suggested that about 1.7% failed the screen and were subject to further scrutiny.

Of these 184 infants, 133 (72%) were subsequently diagnosed either provisionally or finally with 'something' including autism, developmental delay, learning disability, etc. The majority of the 133 diagnosed infants were diagnosed with learning disability (42%) followed by autism (24%).

Never one to say the glass is half-empty, you can imagine that there are a few 'issues' with this scenario which complicate matters slightly; some of which I have touched on in a previous entry. In short: the relative stability of symptoms during a period of quite a lot of developmental change for the infant. If also the headlines read that the screen was pretty good at predicting learning disability (LD) I would say, yes the figures aren't bad at all. Given that autism came quite a distant second to LD, I am slightly hesitant to say that the headlines are indeed so accurate and we can truly spot autism so early and so accurately. Just my deduction from the data supplied (don't shoot the messenger).

What is perhaps more interesting is what the study says about the relationship between autism and LD, and in particular the similarity in early developmental patterns between the two conditions based on the CSBS-DP. LD is perhaps the most frequently cited co-morbidity to autism, followed by epilepsy. Further details on the various criteria and definitions used to diagnose learning disability can be viewed here (see page 26 onwards - apologies for some of the terminology used).

What is inferred is that the developmental course of LD and autism at 12 months share some significant overlap. Speculation: does this mean that autism and LD at 12 months are one and the same which depart from each other as development continues post 12 months? If so, what factors determine this dispersion of conditions? I assume that some follow-up to the Pierce study will be forthcoming in subsequent years. Only then when the mist of development starts to clear, will we get a more accurate picture of the success of the CSBS-DP for predicting autism at such an early age.