Back from the far-away land of deadlines with some very-belated information from IMFAR (International Meeting for Autism Research) 2007, and some thoughts about researching autism interventions.
I saw two presentations involving data from early ABA-based interventions at IMFAR this year. Both studies are community-based.
A Canadian study was presented by Isabel Smith of Dalhousie (the abstract is here). Her ongoing study involves children in a government-funded ABA program in Nova Scotia. She was presenting data after one year for 27 autistic children with an average age just over 4yrs at intake, 6 of whom are described as "non-verbal".
Dr Smith reported improvements in various measures, 6 and 12 months after the children started in their ABA program. Does this mean the chosen ABA program is "effective"?
It looks like we'll never find out, because there is no intention of ever having a control group. Dr Smith cited ethical considerations as to why a control group was unthinkable.
That's a familiar argument: ABA-based interventions are known to be effective, so denying any autistic children the benefit of an ABA program is unethical. This is a subject in itself which I hope to get back to.
In this case, there's also the greater context: Dr Smith's study is an offshoot of the Nova Scotia government's decision to provide ABA-based interventions to autistic children in NS. First, the decision was made to fund the intervention (that is, provide a public program). Then the idea was to study it, once it was in place, to see if it worked.
So assigning children to a group that would be denied what had already become a government-provided program was out of the question. So far as I can tell, the utility of a non-randomized controlled trial (allowing for parents who might choose a different program for their autistic children) was discarded. Ergo--the effectiveness of this government-funded ABA program will be determined solely by an uncontrolled trial. This is now a CIHR-funded study.
And it all sounds like great news for autism advocates. ABA programs are indeed considered so effective--by governments, researchers, the CIHR, etc.--that controlled trials comparing the effectiveness of an ABA program to another program are rejected as unethical.
But the ABA program being funded by the NS government, and being studied in Dr Smith's uncontrolled trial, is a non-intensive, 15hrs/wk PRT (Pivotal Response Training) program. While PRT has some scientific support, there has never been a major controlled trial of PRT as a comprehensive autism intervention. So there has never been a major controlled trial of PRT as it is being studied and funded in NS. Regardless, non-intensive PRT is being assumed to be effective as a comprehensive intervention, in the absence of any evidence for this, such that any experimental design that risks being informative about its effectiveness is considered unethical.
Dr Smith's study is a ringing endorsement of the low/no standards of science and ethics that autism advocates have vigorously and successfully demanded for autistics. It is perfect evidence of why, in the presence of a massive bulk of autism intervention research, researchers still cannot scientifically claim to know how to help autistic individuals (Volkmar et al., 2004).
The other community-based ABA study was from the UK. It was presented with remarkable efficiency by Patricia Howlin. You can find the abstract here.
This study included an experimental group of 28 preschool children in early intensive ABA programs, and a control group of 16 children in "autism specific nursery provision" which, unlike intensive ABA programs, is fully funded by the UK government. Because this is a community study, the groups were not randomly assigned. Outcomes after one and two years were compared.
Dr Howlin showed graphed data from the experimental group kids first. As a group, they clearly improved over time in the chosen measures--just like the kids in Isabel Smith's study.
Beside these data, she then added the data from the control kids. It looked the same as the ABA kids' data. And indeed, she found no significant difference in outcome measures between groups after two years.
Instead, she found large individual differences among the children, regardless of which group they were in. Dr Howlin presented this data graphically.
Having presented her own data, she went on to present a mini-meta-study (well, I can't think of a better name for it) of ABA group designs. She started with the long list of how these studies are inconsistent with each other in multiple aspects of their design. Then she graphed how widely (and wildly) results in various measures differ among the ABA group designs where these data are available.
Dr Howlin's ABA study is in press at JCPP (Journal of Child Psychology and Psychiatry), but not yet available online. I'd like to see the paper before I comment any more on her actual findings. But if her findings hold up under scrutiny, they will be evidence of how autistics have been ill-served by the widely-accepted and -practiced autism advocacy position that true experimental designs--and more generally the scientific and ethical standards that serve to protect and benefit all non-autistics--are not something autistics deserve.
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.