Monday, February 02, 2009

Revisiting the costs of autism
Michael Ganz of Harvard has published two papers about the costs of autism, a book chapter in 2006 and a peer-reviewed paper in 2007. Media attention followed, including a 2007 Globe and Mail story under the headline "Autism a lifelong burden, study shows."

Dr Ganz's work was embraced by leading autism advocates in Canada, as evidence that unlimited ABA-based autism interventions must be funded by governments, or else staggering costs to society will ensue. Here is leading autism advocate Harold Doherty, writing in 2007 in response to Dr Ganz's work:

"In Canada the courageous parents who litigated the Connor Auton case started telling government over a decade ago that the financial burden to society down the road would be tremendous if early ABA intervention was not provided intensively right away and funded."
Here is Senator Jim Munson, leading autism advocate, as reported by CBC, also in 2007:

"Not providing ABA treatment is more expensive because it involves respite care, group homes and institutionalization, said Munson, who referred to a Harvard University study that put the annual cost of autism to Canada's economy at $3.5 billion."
Senator Munson again, from a 2007 speech:

"A recent study by Harvard University demonstrates that each child in the U.S. with autism will cost society about $3.2 million in medical and non-medical costs over his or her lifetime [1]. With the rising number of diagnoses of autism, the eventual costs will be staggering. We're in a situation where spending on treatment, even expensive treatment, is actually a savings."
The footnote [1] is to Dr Ganz's 2006 book chapter, and the sole treatment referred to by Senator Munson, as "proven to work," is ABA-based intensive intervention.

FEAT in Canada, which is also called "Medicare for Autism Now," is a powerful autism advocacy lobby group. FEAT demands that Lovaas-type ABA be mandated as medically necessary treatment for autism under the Canada Health Act. Here is FEAT in 2008:

"AUTISM is a North American health care epidemic of staggering proportions... The costs to the Canadian economy of NOT treating autism is $3.5 billion a year... the cost in human suffering is immeasurable."
While FEAT makes no direct reference to Dr Ganz's work, the $3.5 billion figure does not come from any existing Canadian study. According to the CBC, Senator Munson also used this figure and attributed it to Dr Ganz, and indeed this figure shows up in the Globe and Mail story about Dr Ganz's work that I referred to above. Dr Ganz's published work about the costs of autism does not make any mention of Canada, but Canada's leading autism advocates have apparently embraced an estimate based on Dr Ganz's US work and published only in a newspaper.

In any case, the premise of these leading Canadian autism advocates is the same. First, they use Dr Ganz's work as an occasion to unscientifically deny the existence of older autistics. Like many autism advocacy signature positions, this is an excellent way to ensure that many autistics will have very difficult lives and will be unlikely to have good outcomes. But Dr Ganz's work, if you read it, assumes a high, stable rate of autism.

Second, leading Canadian autism advocates cite Dr Ganz's work as evidence that government funding for ABA-based autism interventions is the only way to avoid the staggering costs of autism.

In the paper reporting his methodology, Dr Ganz does write about the costs of behavioural interventions for autism. From the "Behavioral Therapies" section of Ganz (2006):

"Although it is not completely clear how effective different types of behavioral interventions are for children with autism, it is rather well accepted that some type of intervention should be initiated. Because their use is becoming more pervasive and as more states are legislating that behavioral therapies become covered services as part of health insurance plans, their costs are included here."
Then Dr Ganz goes on to describe the possible benefits of these interventions:

"However, as the estimates of effectiveness, and hence the financial and nonfinancial benefits, are controversial and because the correspondence between improvement in symptoms and the costs of these interventions are not clear, only the costs of the intervention are enumerated here and are not offset by potential benefits."
Dr Ganz provides a range of references to support the above statements. The upshot is that according to Dr Ganz's model, existing evidence for the effectiveness of behavioural interventions in autism is insufficient for any conclusion that these interventions have benefits, financial or otherwise. Instead, there is only sufficient evidence to conclude that these interventions have costs. This is the model on which Dr Ganz's 2007 paper is based, as is directly stated in the paper.

So autism advocates who promoted Dr Ganz's work about the costs of autism in reality promoted a study whose author concluded that the effectiveness of ABA-based autism interventions has not yet been established, such that no benefits for this kind of intervention can be claimed. As often happens, the question arises as to whether autism advocates read the study and misrepresented it, or did not bother to read it and made public claims about it anyway. Those who are not enamored with autism advocacy standards, and who are genuinely concerned about the future of autistics, will take the trouble to read the primary sources and will draw their own conclusions.


Ganz, M. (2006). The costs of autism. In S.O. Moldin & J.L.R. Rubenstein (Eds.), Understanding Autism (pp. 475-502). Boca Raton FL: Taylor & Francis.

Ganz, M. (2007). The lifetime distribution of the incremental societal costs of autism. Archives of Pediatrics and Adolescent Medicine, 161, 343-349


Unknown said...

I see you are continuing your battle to prevent Canada's autistic children from receiving the benefits of ABA as summarized by the American Academy of Pediatrics in 2007:

The effectiveness of ABA-based intervention in ASDs has been well documented through 5 decades of research by using single-subject methodology21,25,27,28 and in controlled studies of comprehensive early intensive behavioral intervention programs in university and community settings.29–40 Children who receive early intensive behavioral treatment have been shown to make substantial, sustained gains in IQ, language, academic performance, and adaptive behavior as well as some measures of social behavior, and their outcomes have been significantly better than those of children in control groups.31–4

American Academy of Pediatrics, Management of Children with Autism Spectrum Disorders

Michelle Dawson said...

In response to Mr Doherty, I don't see a "battle" in this post, perhaps he could indicate where he found one, and also what exactly in this post he factually objects to.

For example, is Mr Doherty objecting to the quotes from himself and other leading autism advocates that I provided, complete with sources?

Or is he objecting to anyone accurately reporting from Dr Ganz's study? This reporting can be verified by taking the trouble to read the relevant study, apparently something Mr Doherty frowns upon.

In response to Mr Doherty's boilerplate, see this and this, and this.

I also wonder whether Mr Doherty believes that the views of Myers et al. (2007), the narrative review he cites, re the use of risperidone should be definitive, such that Jesner et al. (2007) and Tyrer et al. (2008) should be disregarded.

Alyric said...

Interesting that Gantz 2007 gives a range for the US cohort cost of between 13 billion and 76 billion depending on the assumptions of the model. This stuff isn't etched in stone, it's a model with all the reliability problems of models notably; garbage in equals garbage out.

Anonymous said...

Well, in the spirit of helping the economy, I'll take 3.0 (not 3.2!) million from the feds (they can keep the 200k). I hope they won't take taxes, since I am in fact giving them 200k they would have to spend if I didn't take it.

In exchange, I'll give up my right to any and all government funded autism-related services. I'll also pay all my medical bills, for life, in cash.

They can even pay it out over 20 years or so as an annuity, although I think with the need to "stimulate" the economy, it would perhaps be better if I got the money up front so that I could put that money to use - and create jobs. ;)

I'll bet with 3 million dollars, I'll have a lot better life than if I received every possible autism service that is currently provided. Certainly it would be better than a residential option paid by the government. And I'll save money, and I won't just toss that money into this imaginary incinerator that autism is when it comes to economic decisions (never mind that this money predominantely pays salaries, a very good use of money for the economy, not a bad one as it is characterized when saying autism "costs" society).

Michelle Dawson said...

While we're all waiting patiently for Mr Doherty to explain what, factually, he is objecting to in the original post (information about the standards Mr Doherty wants applied to autistics can be found here and here) ...

Here is a post about a Globe and Mail story about the costs of ADHD. Compare it to the Globe and Mail's costs-of-autism story, spot the difference, and consider (this being one small example among an avalanche of others) the high costs of autism advocacy.

Anonymous said...

The effectiveness of ABA-based intervention in ASDs has been well documented through 5 decades of research by using single-subject methodolog

With all due respect to the AAP, that's basically an oxymoron. It's like saying that the effectiveness of Acupuncture has been well documented by case reports.

About the costs of autism, they are high, and I'd argue that's a self-fulfilling prophecy for the most part. It could be demonstrated that what is spent on autistic children depends on the belief system of the parents and on how affluent the parents are.

KeithABA said...

"With all due respect to the AAP, that's basically an oxymoron. It's like saying that the effectiveness of Acupuncture has been well documented by case reports."

So in your view, the methodology of single subject designs are the same as the methodology of case studies?

Michelle Dawson said...

My suggestion is to actually read the references provided in Mr Doherty's currently favoured paragraph, re single subject designs and what are described as (but aren't necessarily) "controlled studies."

For an example of an ABA single subject design, see this classic paper, published in the flagship ABA journal.

KeithABA said...

In the linked article, the methods are clearly very different than you would see in a case study. I'm not sure how that example relates to Joseph's comment equating single subject design to case study.

We can discuss ethics of this study seperately from a discussion of the methodology of the study.

Anonymous said...

Single-subject design suffers from the same major issues case reports suffer from:

1. We just don't know about single-subject trials that failed. It's unlikely they would be published. It's not surprising that the single-subject trials you find in the literature are almost always successful (as are the Secretin case reports, for example).

2. There is no control group. The subject is their own control.

3. There is no way to control for subjectivity in assessments.

4. There is no way to control for the placebo effect.

5. There is no way to control for coincidental remission (this is relevant in relation to #1).

Single-subject designs are only slightly better than case reports in that they are more systematic. However, there's no reason there couldn't be very systematic case reports.

Michelle Dawson said...

In response to Joseph, that's a good list, but many other things could be added, including the problem of poor to non-existent standards of diagnosis, and the weakness of reversal designs (see Lord et al., 2005).

Also, at least in Canada (and elsewhere also), autism advocates are not lobbying for non-intensive single-behaviour interventions. They are lobbying for comprehensive early (and carrying indefinitely on) intensive ABA-based interventions, as public policy.

The ABA single subject designs have not, according to the (overwhelmingly poor quality) ABA group designs, contributed to better outcomes for autistics in the kind of comprehensive ABA-based interventions autism advocates are demanding--not in decades and across hundreds (or is it thousands) of presumably successful single subject designs.

Anonymous said...

BTW, in evidence-based medicine, single-subject designs would apparently be Level-II-3. From Wikipedia:

Level II-3: Evidence obtained from multiple time series with or without the intervention. Dramatic results in uncontrolled trials might also be regarded as this type of evidence.

This is lower than case-control studies and non-controlled studies. (I can find you some successful non-controlled studies of Secretin).

Case reports are the very next level, Level-III:

Level III: Opinions of respected authorities, based on clinical experience, descriptive studies, or reports of expert committees.

Incidentally, Level III evidence is the type of evidence Harold Doherty seems to prefer.

KeithABA said...

This argument can go on forever, but the bottom line is they are different tools to investigate a hypothesis. Both group and single subject designs have weaknesses. Replication in single subject design is very important, as single subject designs have less external validity. I'm not going to start going point by point, because it already happened here:

I have yet to see a single secretin trial with objective definitions of target behaviors. Repeated measures of these target behaviors, interobserver agreement for at least 80% of the data, or withdrawal of the treatment to establish a functional relationship between the dependent variable and independent variable.

Instead they were AB designs, the weakest of single subject design. You cannot equate the methodology of those studies to those single subject designs published in JABA and other journals.

Michelle Dawson said...

In response to Keith ABA, Lord et al. (2005) wrote, about many ABA single subject designs in autism:

"although reversal designs are easily interpreted, they have had limited value for many of the developmental outcome goals addressed in autism."

Re "objective definitions of target behaviors" see the classic ABA single subject design I referred to above, to get the idea.

Also, Keith ABA does not address the problem of the demonstrated failure of hundreds (or thousands, or whatever it is) of ABA single subject designs to contribute at all to better outcomes (according to the usual standards, and as found via the major ABA group designs) in the comprehensive ABA intervention programs currently being demanded by autism advocates.

And as Ben Goldacre (referring to data published in the BMJ) has noted:

"Well now, in 1995, only 1% of all articles published in alternative medicine journals gave a negative result. The most recent figure is 5% negative. This is very, very low."

Anonymous said...

Both group and single subject designs have weaknesses.

Sure. I could extend that argument and say "both randomized trials and case reports have weaknesses" which is true, but meaningless. There's a reason why there are different levels of evidence. Not all evidence is equally strong.

Now, in my comparison with acupuncture, I see a major problem for ABA, which I'd summarize as follows.

1. Acupuncture appears to be an elaborate placebo. Even though this is not completely resolved, that's where the evidence tends to point.

2. You cannot make a case that the body of evidence of ABA is superior to that of acupuncture. Acupuncture has randomized trials and even some placebo-controlled trials (they use "sham acupuncture"), some positive, some negative.

In other words, you need better evidence to be convincing.

The only area where ABA beats acupuncture is in plausibility - maybe.

The other major problem is the huge difference that exists between the results of non-randomized ABA trials and the one randomized trial. How is this difference explained? It's typical of things like Homeopathy to behave this way.

I wish there were more randomized trials of different quality levels, so we could make a graph of study quality vs. effect. But even given what's there, it's difficult not to be completely skeptical of ABA, as I would be of any made up treatment like GFCF, MB-12 and so forth.

Anonymous said...

It is well worth revisiting the costs of autism. I have yet to see a satisactory explanation of the disparity between Ganz's $3.2 million lifetime costs for autism and the much lower figures from the CDC for mental retardation ($1,014,000), cerebral palsy ($921,000), hearing loss ($383,000)and visual impairment ($601,000).

Indeed, if you add them together, a deaf, blind, retarded person with cerebral palsy works out cheaper than a single autistic person ($2,019,000).

Anonymous said...
This comment has been removed by a blog administrator.
Michelle Dawson said...

The deleted post from Konnie was (you guessed it) yet more badly-placed spam from an autism service provider.

Michelle Dawson said...

In response to Mike S, I agree that Dr Ganz's figures need a closer look. Along these lines, the possibly (and plausibly) very high costs of autism advocacy have to be considered. There is no equivalent of autism advocacy dominating other disability areas.