Autism Treatments: Science, CAM, or None of the Above?

There are plenty of non-evidence-based treatments aimed at ameliorating autism. Holding therapy was once used for this purpose, and may still be used by some practitioners. Chelation therapy is known to be not only ineffective but potentially harmful to the child. But what about the most famous treatment, Applied Behavior Analysis (ABA)? And what about the new treatment on the block, Developmental, Individual-difference, Relationship-based therapy (DIR), or Floor Time, based on the approach of the late Dr. Stanley Greenspan? ABA is often stated to be scientifically supported, even “proven”. DIR, on the other hand, has been included as a complementary and alternative treatment by Lisa Kurtz in her book Understanding controversial therapies for children with autism, attention deficit disorder, and other learning disabilities (Jessica Kingsley Publishers, 2007). The CAM designation generally means that a treatment lacks rigorous research support, and may indicate that the treatment is not plausible in terms of orthodox thinking about psychology or medicine.

ABA is based on behavioristic views of learning, which are well substantiated for both human and animal learning, and for all periods in the human lifespan. This treatment is related to operant conditioning, a method that involves reward or reinforcement of desired behaviors like talking, and the prevention of reinforcement of undesired behaviors like the autistic child’s tendency to flap hands or to become fascinated with objects. (At one time, ABA used punishment as well, but as this seemed ineffective it has been dropped from the method.) Operant conditioning is well-known to change behaviors effectively when the behavior chosen occurs from time to time and when a suitable reinforcer can be identified. It is less easy, but possible, for ABA to encourage a behavior which rarely or never occurs in the desired form; in order to do this, ABA specialists use a method called “shaping” in which they initially reinforce related behaviors (like making sounds with the mouth) and gradually limit reinforcement to the desired behavior (for instance, saying a word). ABA thus has a foundation that is plausible in terms of conventional understanding of human functioning.

There’s plenty of evidence that operant conditioning can alter behavior according to well-understood rules. But is there strong evidence that ABA successfully moves autistic children toward significantly more age-typical behavior? Some years ago, Morton Ann Gernsbacher discussed this issue in some detail (Gernsbacher, M.A.[2003]. Is one style of early behavioral treatment for autistic children “scientifically proven”? Journal of Developmental and Learning Disorders, 7, 19-25). In her article, Gernsbacher pointed out that the 1987 work of Ivar Lovaas, the developer of ABA, which reported a significant effect of the treatment, in fact failed to use a design that is needed for the highest category of research support for an intervention. Lovaas’ study did not assign participants randomly to an ABA treatment group or to another type of treatment. Instead, assignment to groups depended on the availability of a therapist, a method which brings in unknown confounding variables like the effect of holidays. The few similarly-designed, good-quality studies also had assignment to groups determined by practical factors rather than true random assignment. As a result, the strong effects those researchers reported need to be interpreted cautiously. A study in 2000 randomized participants, but reported much weaker positive effects than had come from the non-randomized studies (Smith, T., Groen,A.D., & Wynn, J.W. [2000]. Randomized trial of intensive early intervention for children with pervasive developmental disorder. American Journal of Mental Retardation, 105, 269-285).

As a result of these facts, we can conclude that although it may be appropriate to say that ABA is the best-supported method for treatment of autism, it is not appropriate to speak of it as “scientifically proven” or even as evidence-based at the highest level. To the best of my knowledge, no adverse events attributable to ABA have been reported, so this treatment does not belong in the “potentially harmful” or “of concern” category.

What about DIR? Unlike ABA, DIR is not based primarily on principles derived from empirical research, but instead has its foundations in a theory derived by Stanley Greenspan from a variety of sources: psychoanalysis, Piagetian theory of learning, the sensory integration theory of Jean Ayres, and studies of language development. DIR emphasizes an essential connection between emotional development and motivation to communicate with other people, and the capacity for language, thought, and problem-solving. This method focuses on the normal sequence of developmental change, on the formation of social relationships, and on individual differences between children. DIR is thus plausible in terms of commonly accepted ideas about developmental change, although its greater complexity of sources makes it less likely to have the type of support that is foundational for the simpler ABA approach.

Like practitioners of many other treatments for young children, DIR advocates have done little so far to test the efficacy of their preferred intervention. Devin Casenhiser, a psycholinguist in Toronto, has been carrying out a randomized trial study of DIR, but has apparently not yet published it. A number of doctoral dissertations have examined various effects of DIR, including attitudes of parents and of teachers about its use, but have not systematically investigated child outcomes.

It would be impossible to conclude that DIR is evidence-based when so little has been done to examine the outcomes of the treatment. As usual, however, we have no way of knowing whether there are unpublished studies with negative results, or even with the conclusion that DIR had a worse effect than another treatment used as a comparison. Like ABA, DIR appears to have been without reported adverse events.

Should we, then, agree with Lisa Kurtz and describe DIR as a complementary and alternative treatment? I confess that I was shocked to see this classification, as so much of the background of the treatment is highly plausible and conventional in nature. I would see DIR as an little-examined intervention of a conventional type-- a genuinely “experimental” treatment-- as opposed to CAM treatments that are not only without a basis in evidence, but are implausible and incongruent with existing information about development. It may be that plausibility is as important as empirical evidence in determining whether a treatment belongs to the CAM group, but in many ways plausibility is more difficult to determine than evidence, as it requires close examination of the premises and reasoning foundational to a treatment, rather than the simpler analysis of research designs and statistics.

It seems that neither statement about these methods for treating autism is clearly substantiated. ABA has much weaker scientific support than is often claimed. DIR, which makes few claims for research support, may not be correctly designated as complementary and alternative in nature, but I wish DIR proponents would comment on the inclusion of their method in the Kurtz book. I’d like to know whether they too accept the CAM categorization.