Screening for Autism

I have a couple of problems with the AAP's new screening proposal for autism.

First, the recommendations, to be published in November, appear to be centered on the developmental exam. Is it really screening, if it is part of the routine exam? Our purpose with the frequent schedule of visits is to pick up any number of abnormalities, many of which do not really qualify as screening, in the usual sense. Screening, as a term, should be reserved for specific testing for specific conditions.

Symptoms that raise concern include not smiling at the sound of a parent's voice by the age of 4 months and the loss of language or social skills at any age. These are non-specific, that is, although they are common in autistic children, there are other causes and may be normal in non-autistic children. Babies who don't babble by 9 months or point by 1 year may have other developmental problems, or in fact, none at all.

Rigorous assessment of a screening recommendation also needs to identify certain statistical properties of the test. Most physicians are familiar with the concept of sensitivity and specificity, but once loosed on an unsuspecting world, the predictive value of a test is more important. It's not about what proportion you will miss, or what proportion of your patients has another condition. It is about how often your positive is really positive and how often it is simply wrong.

We never argue the benefit of a timely diagnosis, but doesn't the cost of positive screen for something as emotionally charged as autism make it seem at least the slightest bit unwise? Not to mention that, in fact, treatment is poor. The possibility that early intervention improves outcomes is not based on hugely significant outcomes. As a parent, you're still stuck with an autistic kid; you just knew about it a little sooner.

In the absence of definitive and significant evidence of the cost-effectiveness of an intervention, there should be no screening. I am afraid that the cost of false positive simply outweigh any potential benefit.

The fact is, we still examine children regularly and attempt to detect developmental delays including autism as soon as we can. My problem is that they are calling it a screening test, which it is not. My problem is they will unleash this concept on an unsuspecting public with little apparent consideration from the rest of the medical community. It was published in the Washington Post before a medical journal. Poor form, but then, who cares, right? It's all entertainment anyway.