Detecting Autism at Birth

An occasionally-made claim by the anti-vax movement is that autism cannot possibly be genetic because we don't see it at birth.  Anyone who has a background in developmental biology is rolling their eyes after reading that, but to many it seems a logical argument.

It isn't, because humans are born with an incompletely developed brain, and as such developmental abnormalities can occur after birth.

But that has little to do with this particular claim, as autism can be diagnosed under laboratory conditions at/near birth.  Hence, this claim of the anti-vaxers is yet another of their lies (and boy, do they have a lot of them).

Now one may ask, if autism is detectable at birth, why is it in the real world it isn't detected until 18 months or older?  The reason is pretty simple, and I'm going to use two older papers as examples of why this is the case.  Both are available for FREE at the links below:


Autism detected AT BIRTH, upto 6mo:
PNAS November 10, 1998 vol. 95 no. 23 13982-13987
Aspergers detected in infancy:
PNAS August 10, 2004 vol. 101 no. 32 11909-11914

In both of these papers autism is detected young - from birth to 6mo in the first paper, and aspergers  from 3mo onwards in the second.  The later is particularity interesting, as aspergers is a "mild" form of autism often not detected until the age of 5 or older.

If you look at either of those papers you'll see why autism is not often detected in the very young; to detect autism at these early ages requires hours of analysing the movement of these children.  On the right I have one such example, of a plotting system used to analyse the movement of arms and legs.

In addition to using rigid analysis of movement, both papers also used pre-set criteria to ensure unbiased measurements.  On such example is shown below (on the left) of such a criteria; in particular, the movement pattern of an infant moving from its back to front.

Finally, they looked at facial movements, in particular looking for movements typical of older autistic children.

The conclusion in both papers was clear - the symptoms of autism could be identified in children before six months of age with great reliability.

So why don't clinicians diagnose autism at a younger age?

The answer is simple - the diagnosis of autism at these young ages requires in-depth analysis of hours of video recordings.  Even with modern technology, our ability to automate this kind of analysis is very limited.  Furthermore, even utilising modern motion-capture methods would require specialised (and very expensive) studios and software.  Finally, we'd have to screen each and every child to detect autism, for the simple reason that its unlikely that most parents would recognise the subtle symptoms which occur in infants and seek medical advice.

So contrary to anti-vax claims, autism IS detectable at birth.  Unfortunately, a usable and affordable clinical test is a long ways off - in fact, we'll probably have a genetic test long before we have a motion-based one.