Friday, August 03, 2007

Autism and ABA in the UK: A controlled trial

The ABA non-randomized controlled trial presented by Patricia Howlin at the International Meeting for Autism Research this year has recently been made available online. You can find the abstract of this in press paper here [edit: If you follow the link, you'll see that this paper has now been published]:

Magiati, I., Charman, T., & Howlin, P. (in press). A two-year prospective follow-up study of community-based early intensive behavioural intervention and specialist nursery provision for children with autism spectrum disorders. Journal of Child Psychology and Psychiatry.
As I wrote earlier, Dr Howlin's study is a prospective, community-based study comparing the outcomes of 28 autistic children receiving home-based intensive ABA-based interventions (the ABA group) with a control group of 16 autistic children receiving "autism specific nursery provision" (the nursery group). The interventions are described as being "typical" of ABA and nursery services available in the UK. The children were between 23 and 54 months and had a very wide range of IQs (16 to 138) at intake. They were followed up after ~26 months.

The ABA group received intensive 1:1 Lovaas-type ABA, with two children receiving additional Verbal Behaviour services. For 20 ABA group children, ABA services were provided by "recognised ABA organisations in the UK, Norway or USA", while the remaining 8 children had independent providers. The nursery group was distributed across 10 different schools. They received what would be called "eclectic" intervention, where the "most common named practices" included TEACCH, PECS (an ABA-based approach), Makaton, and SPELL (from a 2001 guide published by the UK National Autistic Society), as well as "other developmental and behavioural teaching methods".

And--as reported in the abstract--after a bit more than 2 years, there were no significant differences to report between the ABA and nursery groups in a multitude of outcome measures. On only one measure was a difference approaching significance found in favour of the ABA group (Vineland Daily Living Skills, where p=.06). In contrast, there were major individual differences among the children in progress made, regardless of which group the children were in.

As happens in non-randomized controlled trials, the two groups were not matched on many variables at intake. The authors accounted for only one of these unmatched variables--a significant difference in mean intake IQ which favoured the ABA group (83 for the ABA group; 65 for the nursery group). Other significant differences at intake (e.g., Vineland socialization, Vineland composite, parental education) favoured the ABA group and were not accounted for, while other unaccounted for differences favouring the ABA group were just short of being significant (e.g., the ABA group was younger at intake). None of the between-group differences at intake favoured the nursery group.

The authors also made no attempt to account for gender. The nursery group had a higher percentage (33%) of females than the ABA group (4%). In my view (and apparently, in the authors' view), this isn't a major issue and if anything--according to the lore that autistic females are at a disadvantage--this difference would again favour the ABA group.

The ABA group also had the advantage of receiving a significantly more intensive intervention than the nursery group, both at the outset of treatment and at follow-up, with the ABA group at ~33hrs/wk (range 18-40) and the nursery group at ~26hrs/wk (range 15-30). Within each group, intensity of intervention did not change from the outset to follow-up. That is, the children's need for services did not decrease over the course of about two years. The nursery group received little in the way of 1:1 intervention (6hrs/wk average, compared to the ABA group, where the full ~33hrs/wk was 1:1).

Because they received a more intensive intervention, the ABA group also received a significantly higher total number of intervention hours than did the nursery group--an average of 3415 hours versus 2266 hours per child. This is a difference of 1149 hours favouring the ABA group.

Using the average intensity of intervention for the ABA group, the ABA group received an equivalent of 35 weeks more in the way of intervention time than the nursery group children. Using the average intensity of intervention received by the nursery group, the nursery group received an equivalent of 44 weeks less in the way of intervention time than the ABA group children. That is, for the amount of intervention to be equivalent in both groups, the nursery group children would have had to receive an additional 44 weeks of intervention (at their average intensity of intervention).

On the other hand, this study found no relationship whatsoever between intensity of intervention and any outcome measure. This is keeping in mind that the range of intensity in the ABA group was 18-40hrs/wk. In my IMFAR poster this year, I mentioned the failure of the ABA literature to relate higher intensity of ABA-based interventions with better outcomes, and in Dr Howlin's study, that failure continues.

Another factor that had no effect on outcomes was age at intake. Whether the children were younger or older at intake, across the range from 23 to 54 months, made no difference to how well they did. The popular if not ubiquitous contention that, when it comes to autism interventions, "earlier is better" has failed--again--to be supported by evidence from a controlled trial (for a previous failure, right up to intake at age 7, see Eikeseth et al., 2002, 2007).

And regardless of the ABA group starting with an average IQ well within the "high-functioning" range (indeed, the intake IQ here for the ABA group is the same as the follow-up IQ in Lovaas, 1987), after ~2 years of intensive ABA, all children in this group were still receiving ABA services and none was in a mainstream school without 1:1 assistance. This is also regardless that almost one-third of the ABA group did not have the specific diagnosis of autism at intake, instead being assigned a subthreshold "ASD" diagnosis.

On average, neither group did well. They both made progress in age-equivalent scores but, as the abstract reports, standard scores changed little. Individuals either did well or did poorly, and this was unrelated to which kind of intervention they received, the intensity of intervention, or the age at which intervention began. This is similar to what was found in a recent Canadian observational study (Eaves & Ho, 2004).

Dr Howlin's study, like all studies, has weaknesses and limitations (and of course I want a lot more information about practically everything), some of which are addressed by the authors. I find it ironic that the authors argue for autism-specialized services by citing Cohen et al. (2006). While it is difficult to compare across studies, it could be argued that the control group in Cohen et al. (2006), which received completely inadequate services that no one should recommend (generic segregated special education of low intensity), fared better by the usual standards than both groups--the ABA group and the nursery group--receiving autism-specialized services in Dr Howlin's study. Dr Howlin and colleagues also fail to point out that when unmatched intake variables are accounted for in Cohen et al. (2006), the few significant differences between groups (ABA vs generic segregated special education) in all but one outcome measure (classroom placement) disappear.

Dr Howlin and colleagues conclude that:

Our data support the growing consensus that no one intervention for children with ASD is universally superior to all others (NIASA, 2003).
My own conclusion is one I've arrived at before: after more than 60 years of autism research, encompassing the colossal existing autism intervention literature, researchers still cannot scientifically claim to know how to help autistic individuals (Volkmar et al., 2004). The best adult outcomes reported in the autism literature continue to belong to individuals who grew up before the current era of early interventions and who as children met the narrowest, strictest autism diagnostic criteria ever devised (Dawson et al., in press).

I'm prepared to agree with Dr Howlin that autism-specific services are important. But even the evidence Dr Howlin and colleagues drum up in support of these services glaringly reveals how poorly autism research has served the interests of autistics. Even if you leave out ethical concerns that would be paramount with any non-autistic population, the major currently-popularized and -marketed autism educational interventions (as opposed to non-popularized non-marketed science-based approaches; see Aldred et al., 2004; Gernsbacher, 2006) leave little to choose from. Autistics deserve a whole lot better.


Aldred, C., Green, J., & Adams, C. (2004). A new social communication intervention for children with autism: pilot randomised controlled treatment study suggesting effectiveness. Journal of Child Psychology and Psychiatry, 45, 1420-1430.

Cohen, H., Amerine-Dickens, M.S., & Smith, T. (2006). Early Intensive Behavioral Treatment: Replication of the UCLA model in a community setting. Journal of Developmental and Behavioral Pediatrics, 27 (S2), 145–155.

Dawson, M., Mottron, L., & Gernsbacher, M. A. (in press). Learning in autism. In J. H. Byrne (Series Ed.) & H. Roediger (Vol. Ed.), Learning and memory: A comprehensive reference: Cognitive psychology. New York: Elsevier.

Eaves, L.C., & Ho, H.H. (2004). The very early identification of autism: outcome to age 4 1/2-5. Journal of Autism and Developmental Disorders, 34, 367-378.

Eikeseth, S., Smith, T., Jahr, E., & Eldevik, S. (2002). Intensive behavioral treatment at school for 4- to 7-year-old children with autism: A one-year comparison controlled study. Behavior Modification, 26, 49–68.

Eikeseth, S., Smith, T., Jahr, E., & Eldevik, S. (2007). Outcome for children with autism who began intensive behavioral treatment between ages 4 and 7: A comparison controlled study. Behavior Modification, 31, 264-278.

Gernsbacher, M.A. (2006). Toward a behavior of reciprocity. Journal of Developmental Processes, 1, 139-152.

Lovaas, O.I. (1987). Behavioral treatment and normal educational and intellectual functioning in young autistic children. Journal of Consulting and Clinical Psychology, 55, 3–9.

Volkmar, F.R., Lord, C., Bailey, A., Schultz, R.T., & Klin, A. (2004). Autism and pervasive developmental disorders. Journal of Child Psychology and Psychiatry, 45, 135-170.