Tuesday, August 4, 2009

Management of Children With Autism Spectrum Disorders

Every so often, I find a research article that makes me want to celebrate. On the other hand, I also periodically find articles that make me want to slam my head against a wall in response to their sheer stupidity.

And, although it's rarer, I do occasionally find an article that makes me want to do both. For obvious reasons, it's almost inevitably different sections of the same article that do each. I haven't had the experience of having both reactions to the same sentence... or the same paragraph, for that matter.

I bring this up because that last is the sort of reaction that I had to this article (also findable here), which was published two years ago in the absurdly prestegious journal Pediatrics.

Entitled "Management of Children With Autism Spectrum Disorders", the article is basically a comprehensive introduction to autism, intended for general practitioners so that they can familiarize themselves with the field for whenever they get an autistic patient.

It's a nice idea -- and the journal it got published in is probably the one which pediatricians are most likely to read. Unfortunately, a good bit of the information contained within is biased or just plain wrong.

Now, to be fair, the article is two years old and a lot of the studies refuting points in it were published in the last two years (thanks, Michelle!). Some of it, however, was just plain wrong, even with what we knew then... and to say that the clinical philosophy statement is questionable is a major understatement.

To quote:

The primary goals of treatment are to minimize the core features and associated deficits, maximize functional independence and quality of life, and alleviate family distress. (pp. 1162-1163)

The first clause of this, "to minimize the core features and associated deficts" comes in direct conflict with the remainder for the simple reason that autism isn't just a set of deficits. "Minimizing the core features" of autism includes those srengths as well as weaknesses -- it's trying to turn an autistic person (to the degree possible) into a non-autistic person. This is pretty thoroughly appalling... but almost to be expected.

The article also drags up the whole theory of mind myth, the joint attention thing, the myth of the "need" for intensive behavioral intervention... the list goes on and on. In general, it drags up a crapload of invented, so-called "knowledge" and passes it off as truth.

Hell, it even manages to miss the fact that autistic children learn and develop in ways completely out of line with neurotypical developmental timetables and skill hierarchies.

The worst section of the paper, however, is the section on applied behavior analysis. The errors start with what ABA is... and just keep piling up from there. If you believe the article, ABA is:
... the process of applying interventions that are based on the principles of learning derived from experimental psychology research to systematically change behavior and to demonstrate that the interventions used are responsible for the observable improvement in behavior (p. 1164).

... umm, no.

Cooper, Heron, & Heward's (2007) textbook, Applied Behavior Analysis (2nd Edition) is pretty much considered (for the time being, anyway) the definitive reference on ABA. Its definition is:
Applied behavior analysis is the science in which tactics derived from the priniciples of behavior are applied systematically to improve socially significant behavior and experimentation is used to identify the variables responsible for behavior change (p. 20).

In other words, Myers et al. are wrong on two major points:
  • ABA is a science. The process of using tactics derived from that science is technically referred to as clinical practice based on ABA. While this may seem like quibbling, it's actually a fairly important distinction. Like any science, ABA offers predictions -- "what will happen if I do this?" -- theories, and very little else... although there are certainly interventions derived from it and experimental tests of those interventions (with the tests falling under the "ABA" category, but the manuals arguably not).
  • Practice based on ABA does not use "the principles of learning derived from experimental psychology research". ABA-based methods use the principles of behavior derived from EAB and ABA research. While some of these can be described as (or can act as) principles of learning, EAB and ABA research is only a subset of experimental psychology research... and much of the non-behavioral research has been devoted to learning. Simply put, Myers et al.'s definition suffers from both deficiency and contamination.

I could go on to critique the other various errors in this section... but won't. Frankly, the fact that the article couldn't even get the definition of ABA right about sums up the accuracy of the article's ABA section... and makes me wonder how the heck it got published in Pediatrics of all places!

While I'm not overly familiar with the developmental treatment models, there are a number of issues with the article's section of them as well... although, to be fair, some of these may be flaws in the actual models themselves. The DIR model, for instance, "targets the core impairment in social reciprocity" (p. 1165) despite the fact that while such a deficit does exist, it certainly doesn't lie with autistic children!

The section on speech and language therapy is comparatively good (which shouldn't be confused with an unqualified "good"), but the recommendations the article makes regarding social skills instruction are, frankly, abominable from an etical standpoint (not to mention that it drags up certain fallacies regarding joint attention yet again).

I'm not even going to get into the section on comparative efficacy. Let's just say that it's at least better than the section on ABA... which means that it at least got the definition of its subject matter right.

The section on sleep disturbance, however, was just plain scary. Benzodiazepines? Zolpidem? Trazodone?!? By contrast, the reccommendation of zaleplon was almost understandable.

Its section on psychopharmacology wasn't much better.

There was also one line which made me blink and react in a manner fully consistent with typical usage of the acronym WTF. Specifically: "However, if a child with an ASD presents with symptoms such as chronic or recurrent abdominal pain, vomiting, diarrhea, or constipation, it is reasonable to evaluate the gastrointestinal tract."

This needed to be said?

After all of that, however, the article had some truly outstanding points. Its coverage of complementary and alternative medicine (AKA "woo") is excellent (barring, perhaps, its coverage of facilitated communication, which I can't comment too much on). One line that comes to mind as particularly noteworthy is:

Some treatments, such as intravenous chelation, may be particularly dangerous and should be discouraged. One child with autism died as a result of chelation with edate disodium... despite the facts that a causal association between mercury and ASDs has not been demonstrated, there is no scientific evidence that chelation is an effective treatment for ASDs, and the effectiveness of chelation therapy to improve nervous system symptoms of chronic mercury toxicity has not been established. (p. 1174, chemical formula of edate disodium omitted)

Another worthy line:

Educational programs should be individualized to address the specific impairments and needed supports while capitalizing on the child's assets rather than being based on a particular diagnostic label. (p. 1167)

In short, I have highly mixed feelings about this article... but at least reading through the contents of the reference section will keep me busy for a while.


  1. Good work, Alexander! :-) I'm looking forward to reading the articles you linked to that I hadn't read yet.

  2. Apart from your comments about the proper definition of ABA do you take issues with conclusions drawn in the AAP article about sustained gains made by autistic children who received ABA intervention? Do you also discredit previous reviews of ABA effectiveness such as the MADSEC Report, the NYSDOH reviews, the US Surgeon General and the various reviews of the Association for Science in Autism Treatment?

    How about the opinions of professionals who actually work with autistic children such as Doreen Granpeesheh of CARD, Alan Harchik of the May Institute, Dr Mickey Keenan of the University of Ulster, former Director of Parents’ Education as Autism Therapists (PEAT)or Dr. Paul McDonnell of Fredericton, New Brunswick, Professor Emeritus (Psychology) and Clinical Psychologist? These people are not grad students, or former postal workers. They have spent decades actually working with autistic children, applying ABA principles to help them make the gains specified in the studies quoted above.

    I respectfully disagree with KWombles. Your article is not very good at all. No offense intended.

  3. Mr. Doherty,

    You seem to forget that I am one of those professionals as well. Just what do you think I'm a grad student in? Have you looked through my blog long enough to notice that my program has a heavy practicum component? There are several other relevant questions I could ask, but I think you get the idea.

    I'm due to finish my program in a week or so, too... so this blog will need a new subtitle pretty soon.

    Now that I've addressed the ad hominem portion of your message, let's focus on the actual content.

    The problem with the ABA section of the article that I address is its constant purveyance information that is incorrect, oversimplified, or otherwise erroneous in some fashion. For instance, it lumps ABA-based interventions together in terms of effectiveness, thereby inappropriately generalizing studies done on one form of intervention to another based solely on the fact that those interventions are based on the same theoretical model. The fallacy involved here is something that I believe should be obvious.

    For obvious reasons, a study on the effectiveness of PECS says little to nothing about the effectiveness of Lovaas-style DTT.

    I also object to the way it describes both compliance and imitation as "foundation skills" without further citation. There's a great deal of emperical evidence that imitation (especially motor imitation) as defined in ABA practice is anything but, and compliance...

    I'm not getting into the ethical issues involved with that in a mere comment. If I ever write about it, I'll try for a peer-reviewed journal.

    Then there's the way the article implies that the only (or even primary) criticisms of DTT are the issue of generalization difficulties and the "naturalness" of the teaching environment.

    I could go on and on. Functional assessment methodologies are described as a singular method despite being anything but. Functional analysis and assessment are confused (admittedly an easy enough mistake to make and one that even I make in colloquial speech, but one that has no place in an edited article submitted to a major journal). Elements of theoretical models are taken as fact without citation... etc., etc.

    As for the previous reviews, I cannot speak about most of the ones you cite. I tend to read peer-reviewed literature, which none of the reviews you mention qualify as (to my knowledge). The MADSEC report, however, is in my queue of things to read through, somewhere after a rather interesting paper by Edelson (2005) on criterion contamination in intelligence tests.

  4. One extremely impressive piece of commercial SPAM deleted (had the author link not linked to a "generic viagra" seller, I would have seriously wondered whether that was actually SPAM or just a somewhat incoherent serious author).