Thursday, May 14, 2009

The autism recovery story
You have seen the autism recovery story in the media. Bloggers have jumped in too. I was at IMFAR 2009 in Chicago and saw the educational symposium presentation by Deborah Fein that was the stated cause of all the excitement.

But in her presentation, as in her abstract, Dr Fein did not associate the findings she reported with any kind or quantity of autism intervention or treatment. When speaking at IMFAR, she expressed doubt that this in fact could be done.

Dr Fein clearly added more information when speaking with the media. She expressed her view that recovery from autism, what Dr Fein calls "optimal outcome," was associated with early intensive ABA-based autism interventions. Like all researchers, she is free to say what she wishes to the media.

However, her statements relating kinds and amounts of intervention to outcomes in autistic children are not supported by any of the data she chose to present at IMFAR 2009, either in her oral presentation or in a series of related posters.

Indeed her study design does not permit any conclusions about effectiveness of interventions, no matter how "effectiveness" is defined. She has not conducted a true experimental design in which well-known sources of bias can to some degree be accounted for and therefore the effects of interventions, their benefits and harms, can usefully be assessed.

In order to claim that conclusions can be drawn about the effectiveness of interventions in Dr Fein's study, as it has been presented, you have to be willing to reject and discard even the most basic standards of science. These are the basic standards that automatically protect and benefit nonautistics, the basic standards that were developed because without them people were harmed.

Two other aspects of what Dr Fein presented at IMFAR 2009 are noteworthy. One is that a major paper in the literature, published not long ago, reported a very high rate of what is now being called recovery from autism. To my knowledge, this paper did not make headlines.

Turner and Stone (2007) were looking at the stability of autism diagnosis at around age two years (mean age of 29 months). They carefully used various combinations within what is considered the gold-standard diagnosis: ADOS, ADI-R, and clinical judgment. And they found that early diagnosis, in their sample of 48 originally autistic children, was very unstable.

By age four (mean age of 53 months), only 53% of children originally given the specific diagnosis of autism retained this specific diagnosis, while only 68% were still on the autistic spectrum. The rest were not. That is a 32% rate of "recovery." Only 40% of children originally diagnosed with PDD-NOS at age two still had an autistic spectrum diagnosis by age four. That would be a 60% rate of "recovery."

The children whose age two diagnoses were unstable were largely still regarded as having difficulties, a finding that echoes Dr Fein's work, but they no longer could be diagnosed as autistic.

Turner and Stone (2007) attempted to associate instability of diagnosis (what is being called "recovery") with interventions received, but no association was found. They report:

All children in the sample received speech therapy, and the majority received additional interventions. There were no significant differences between the Stable and Change groups for the amount of speech therapy, t(46) = .30, p = .77, behavioral therapy, t(46) = .78, p = .44, occupational therapy, t(46) = 1.90, p = .06, special education, t(46) = 1.96, p = .06, or regular preschool, t(46) = .13, p = .90. Of the 6 children who received an average of 20 or more hours of intervention per week, 5 were in the Stable group and 1 was in the Change group.
In other words, few of the 48 originally autistic children in this study received what would be considered "intensive" intervention, and the majority of those who did kept their autism diagnosis.

In contrast, Turner and Stone (2007) found that diagnostic instability was related to earliness of diagnosis: the earlier the diagnosis, the more unstable it was.

The authors go on to discuss what may underlie the diagnostic instability they found while using the best available diagnostic standards. In fact there are many possible explanations. One point they do underline, however, is that given their findings,

...extreme caution must be taken when interpreting intervention findings that suggest ‘cures.’
And of course that would include intervention findings that suggest "recovery." This may be all the more so when diagnosis, and therefore intervention, is very early.

Also noteworthy is Dr Fein's use of the term "optimal outcome," a term she prefers to "recovery." Dr Fein and her colleagues have determined the criteria for optimal outcomes in autism, and these criteria assume that autistics who remain autistic are suboptimal. Remaining autistic is an undesired, unsatisfactory, inferior, suboptimal result.

According to Dr Fein, and those enthusiastically promoting her views, we know what an optimal human being is like, and this optimal human being is not autistic. An optimal outcome is not Stephen Wiltshire or Alex Bain or Daniel Tammet or Derek Paravicini or Hugo Lamoureux or Vernon Smith or Richard Bocherds or Danny Melvin or Tony DeBlois or janet norman-bain or Jessica Park or Temple Grandin or the large proportion of autistics found recently to have exceptional skills (Howlin et al., 2009) or me for that matter.

Indeed, Dr Fein's group has proposed loss of exceptional autistic skills as evidence for optimal outcomes. And given the great support Dr Fein's views have received, we should be in little doubt as to the goals of popularly demanded autism interventions, and the consequences for those of us who for whatever reason remain unrecovered. The decision as to which kind of human being is optimal and which kind is not has been made.


Howlin, P., Goode, S., Hutton, J., & Rutter, M. (2009). Savant skills in autism: psychometric approaches and parental reports. Philosophical Transactions of the Royal Society B: Biological Sciences, 364, 1359-1367.

Turner, L., & Stone, W. (2007). Variability in outcome for children with an ASD diagnosis at age 2 Journal of Child Psychology and Psychiatry, 48 (8), 793-802 DOI: 10.1111/j.1469-7610.2007.01744.x


Anonymous said...

Are you familiar with Eaves & Ho (2004)?

Anonymous said...

It's hard to tell from the abstracts; Dr. Fein's work does not seem to be a follow-up of children first seen when they were younger, is it?

Where does the 10% recovery rate reported in the media come from?

Michelle Dawson said...

If you look around this blog, you will find several mentions of Eaves and Ho (2004; here, here, here, for example...), a study I read when it was first published.

To get an idea of what Dr Fein's study (the one she presented at IMFAR 2009) involves, I suggest listening to the long, unedited TV interview I linked to in the original post.

Anonymous said...

I suggest listening to the long, unedited TV interview I linked to in the original post.

Thanks for the link. It's not clear from the interview what the methodology was, but she says at one point that they looked retrospectively to see if the children had "really" been autistic, diagnosed by experts, and so forth.

That would not be an observational study of a group, some of whom moved off the spectrum and some of whom did not. It would not be possible, in fact, to say if intervention was a variable related to outcome.

Dr. Fein is basically speculating substantially about that.

ChrisB said...

Mind you, any treatment methodology could obviously see the advantage of early intervention with a condition that (more or less, up to a point, caveat caveat) goes away in half the cases by the age of four. It would be quite hard not to make a living out of.

Lisa Jo Rudy said...

This is really pretty fascinating to me. Didn't know about the 54% "instability" of autism diagnoses, but am not at all surprised. I'm guessing this may be why pediatricians are loathe to diagnose kids at such a young age or to recommend evaluation: many kids are slow to develop and show odd behaviors - and then catch up quite rapidly.

It's a real conundrum: diagnose early for best outcomes? Or wait and see, and avoid all the insanity that goes along with the diagnosis?

Lisa (

Michelle Dawson said...

Lisa Jo Rudy wrote, "diagnose early for best outcomes?"

Setting aside the question of what a "best outcome" is... At least with ABA-based interventions, existing evidence (keeping in mind that overall quality of the evidence is poor) is not in the direction of earlier-is-better.

The evidence from controlled trials, wherever the relevant information is provided, is that earlier makes no difference. Age at intake is unrelated to outcomes. See, e.g., Lovaas (1987), Magiati et al. (2007), Eikeseth et al. (2002, 2007), and Eldevik et al. (2006).

Howlin et al. (2009), an attempted systematic review of the ABA group designs, could not find one controlled trial where age at intake was a predictor of outcome.

And it was last year that Sir Michael Rutter was reported to state, at the NAS conference in London, that "there is ‘no convincing evidence that benefits of psychological intervention are contingent on either very early or very intensive intervention.’"

This in fact is an accurate statement, if you seek to verify it by reading the relevant peer-reviewed literature.

Anonymous said...

Basically, there's no convincing evidence that the act of diagnosing early is what makes the child's autism label unstable. I think it's simply harder to predict whether the diagnosis will hold.

A lot of experts recommend pushing the age at diagnosis down for better outcomes. But this is not exactly an evidence-based recommendation.

Imagine they push the diagnosis age down to 6 months of age. It's reasonable to suppose there would be a lot more false positives this way.

farmwifetwo said...

For me recovery isn't just learning to talk, learning to control you environment (social, sensory, spatial) but also executive function. The ability to plan to complete a task from A to B. This is the skill we are still lacking in my eldest son that they are working on at school. At home he's fine, but school he still have difficulty. But, he's still only 9 so, I have expectations of a full "recovery".

If someone has autism and can live independantly (without respite of any kind), have a family, children, job.. then IMO... they are "recovered" and are no longer on the spectrum nor should they claim they are.

I was told by my eldest son's last Dev Ped that his Mother probably has Aspergers.... probably do... but since I do not require services, why would I claim I do, why would I go and get a formal diagnosis and claim I speak for those who have severe PDD such as those on the Hub claim to do. I think that is wrong.

My family Dr says the DSM-IV is so vague you can dx anyone with anything.... I'm starting to believe it. I have hopes V makes a lot of changes to the Autism Spectrum and tightens up the requirements for dx.

I don't think any "therapy" is better than any other. For us, therapy has been speech, OT, and simply daily living with 30min of homeschooling tossed in to help keep up with school. Every moment of every day is an opportunity for learning - learning to climb the ladder and go down the slide, making a sandcastle, reading a book with Mom, putting on our socks, going swimming - and as long as learning takes place IMO it's therapy. Also, skills need to be transferable... and ABA does not teach this... I was told "We do not have a program for that", yet their psychometrist said it needed to be done.


Michelle Dawson said...

In response to farmwife2, three things.

First, if you are going to invoke "severe PDD," you should provide your criteria and the instrument(s) you want to use, and at what age you want this measure or these measures used.

So far the best adult outcomes reported in the peer-reviewed literature belong to autistics diagnosed as children with the strictest, narrowest, most "severe" autism criteria ever devised (Szatmari et al., 1989; Kanner, 1973; Farley et al., in press).

For more information (that you can verify by reading the peer-reviewed literature) about autism "severity," see this comment.

My second point is that if you are going to accuse others of whatever, then provide examples with sources, please.

And the third point: one great achievement of autism advocacy is the common belief that diagnostic criteria can be expediently invented and imposed for political reasons, and that it is fine to diagnose or undiagnose anyone, again for expedient political purposes (e.g., to accuse other people of being frauds). These kinds of rock-bottom standards of science and ethics harm autistics, as they would harm any other group.

Larry Arnold PhD FRSA said...

Here is an extract of something I was preparing to write on my own blog about the artificial concept of recovery

Recovery is only a word, a means of indexing a particular concept which is viewed from a particular perspective not shared by every observer of the same phenomenon.

In order for the concept to make much sense one has to recover from some point of failure or brokenness, (the worst a medical modelist could view it as is unfulfilled neurotypical potential, a form of eccentric development from the percieved (and that perception is mutable in itself) norm) Since Autism is not an illness or a breakdown one cannot technically recover, one can only grow outwards from where one began.

You see autism is not a constant of anything, it is a trajectory, and we progress at different rates to each other amongst a number of dimensions, sometimes further forward on one than another, and not necessarily constantly in an upward direction either.

It is very hard to predict at any given point what the outcome will be, and indeed what will affect it positively or negatively. It is an area that is difficult to research because it needs longitudinal studies, with variables that at least have a semblance of comparability. (which of course is rare given the historical trajectory of whatever at any given time the concensus on what autism is, was)

If I had been born a generation earlier, and in the right place I could easily have fitted into either Kanner’s or Asperger’s study, and if I were born a little later I could have been in the Camberwell or the Gothenburg study, who knows. As it is I am in more than one study somewhere, and at least I have achieved that without my brain ending up in one of Autism Speaks pickle jars so they can say “thats the bit that caused him to dislike us so much”

No my generation is a bit of a lost generation, for although the concept of autism had been kicking around consider that it did not formally enter the Devils Dictionary of the DSMIII till 1980 when I was already 25 years old and ineligible for a diagnosis of “infantile autism” as it was then called. Although at that time Autism and the wider spectrum were being mooted, the Camberwell study was back in 78, it was not till 87 that autism encompassed adults, and not till 94 that the spectrum entered the fossil record as were. It took another 5 years to catch up with me, in order to be diagnosed in my own town, and I was still the first case for my GP.

As I say, there are trajectories, and at some point your developmental trajectory might cross with the historical trajectory of the growth of the concept and it’s dissemination.

Larry Arnold PhD FRSA said...

Continued : ---

At some point people’s trajectories are going to escape the dragnet, for although the growth in childhood and then teen diagnosis has been exponential, there are going to be a lot in my generation who will escape. It has to be said however that I never cease to be surprised in meeting people older than me, who continue to be diagnosed.

I would love to be able to put this visual notion of intersecting trajectories and vectors into the elegant language of algebra, but hey I do not think I have “recovered” that far yet.

You see that is what recovery is, it is not ceasing to become autistic, or indistinguishable from your peers, there must be enough residual material for a DSM conforming fanatic like Baron-Cohen to diagnose by even amongst the glitterati.

It is progress, it is growth, it is a trajectory that is all. A word that means more to some than for others. It signifies that something that was perfect became broken and was fixed again.

For a person who has lost some faculty, for it to be restored is recovery. If you began without the faculty, then how are you recovered, you have merely grown and expanded albeit at a differential rate.

Anonymous said...

So far the best adult outcomes reported in the peer-reviewed literature belong to autistics diagnosed as children with the strictest, narrowest, most "severe" autism criteria ever devised (Szatmari et al., 1989; Kanner, 1973; Farley et al., in press).Pat Howlin (2004) notes something to that effect: "the great increase in educational opportunities for children with autism over the past 30 years has not necessarily resulted in significant improvement in outcome in adulthood."

I read about that in Eaves & Ho (2008), a more recent adult outcome study that included autistics of all levels of functioning. Despite the fact that none of the autistics from the Eaves & Ho group were institutionalized, only 4% are determined to have "very good" outcome. (An additional 17% had "good outcome.") It's not straightforward to make a comparison with Kanner (1972), but the "very good" outcome label might have applied to 13% of the autistics Kanner saw.

None had "very poor" outcome, though, which is good - probably a consequence of the lack of big institutions in BC.

When the parents were asked what they thought was helpful, only one named a specific treatment method. Three mentioned services like preschool or speech therapy. What they found most helpful were specific individuals, such as a teacher.

Larry Arnold PhD FRSA said...

"When the parents were asked what they thought was helpful,"

That is the problem isn't it, hardly what you would call an objective measure!

I don't think you can actually measure anything regarding outcome, you can only take a best guess.

To begin with if you are comparing any longitudinal data you need the same measures at the beginning as you need at the end, and you need to be comparing like with like.

All very well, but with a growing child the question is, is like ever like along the route, and how far can you derive a measure that is both accurate and appropriate both at the beginning and the end of the comparison.

Being as autism is only measurable by observation, you just cannot really do it, it is a best educated guesswork dressed up as science.

Anonymous said...

That is the problem isn't it, hardly what you would call an objective measure!

Yes, I wish they would've done better. But I guess it's difficult to ascertain what exactly the parents did in an adult outcome study like that one.

Rarely do they talk about environmental variables as predictors of outcomes in such studies. Kanner talked about institutionalization and living at home, but it was all very qualitative. Other studies talk about parents who pushed for mainstreaming. That's about it.

Michelle Dawson said...

In response to Joseph, I wrote about Eaves and Ho (2008) in this comment.

By 1996 there were no longer any institutions for developmentally disabled people in BC. The highly respected Canadian autism advocate Michael Lewis, who is in charge of BC's autism society and 1st Vice President of Autism Society Canada, would have to look out-of-province for the institution he promises to put his autistic son into (the moment his ABA program lets up...).

In response to Mr Rex, I also was an adult before "Infantile Autism" appeared in the DSM-III in 1980.

CrouchingOwl said...

This discussion reminds me of something my wife read in a textbook on forensic psychology (US based textbook so many not generally apply). The book described that traditionally police take strategies that are intimidating to witnesses such as interrupting them frequently and forcing arbitrary changes in subject etc... makes it feel like an interrogation to simply make a witness statement. Psychologists with limited effect have clamored for a gentler approach more based on actually letting the witness simply talk and say what they remember without the in your face treatment. Afterwards, they designed studies to examine the accuracy in the witness statements between the two groups and found that their beloved theories hadn't actually impacted the quality of witness statements but only the quantity. Despite that complete failure to demonstrate effectiveness, they still advocate for the gentler techniques because they match their theories better.

Similarly, our theories of child development suggest that early intervention have massive impact, such as a strong conditioning event in a infant can produce life long phobias. So we push for earlier and earlier therapies, and afterward when the studies are more or less neutral about the actual effectiveness of the technique, the scientists keep doing it and advocating for it because at least it fits in with their theories.

The examples to me allows this behavior make a perverse sort of sense. Take it for what its worth. I figure science is applied when we ignore these gaps enough to run with the theories, but only progresses when we go after those gaps aggressively. Ignore the gaps and instead of science you just have a bunch of culturally conditioned hunches. So we hit a balancing act between applying what we have to work with presently and trying to figure out anything more on the subject. Autism has such cultural urgency attached we may easily throw the balance off. Thanks Michelle for the good blog.

Barbara said...

Thanks Michelle for the Turner and Stone (2009) which I hadn't seen. But I've made this exact point in my PhD, that very early diagnosis is liable to be unstable. I used that point in the section I wrote on the deeply flawed Lovaas (1987) 'study'.

I used Kleinman et al (2008), and Turner et al (2006), but this latest Turner is even better with much stronger evidence

Thank you very much. And thanks for the detail on the Math not adding up in the mean IQs on the 1987 paper. I'm poor on numbers and would never have noticed that.It's exceptionally useful.

Barbara said...

Sorry, I see your Turner citation is 2007, not 2009.

Ari Ne'eman said...

What criteria is Fein using to define the concept of "recovery"?

Michelle Dawson said...

In response to Ari Ne'emen, in the original post I provided a link to the abstract of a recent narrative literature review, Helt et al. (2008), in which Dr Fein and her colleagues provide their (current) criteria for "recovery" from autism or, as Dr Fein prefers, for "optimal outcome."

If you have trouble finding this paper through your university, write to me and I'll send it to you.

Ari Ne'eman said...

Thank you, Michelle. I appreciate it.

Jannalou said...


I just want to say something about living independently and executive functioning.

It is perfectly possible to have poor executive functioning and live independently. People with ADHD do it all the time. So do people with autism spectrum disorders.

Seeking a diagnosis can be important for many reasons, the primary one being self-discovery and self-understanding. Having a label that explains why you have problems in certain areas of your life can be a huge relief. It can also give you places to look for the help that you need and ideas for ways in which you can work with or around those difficulties.

There are a number of people on the Hub who are only able to live "independently" due to the assistance of staff and/or friends who come in daily to help them do things as simple as getting dressed and as complex as taking the bus somewhere.