Tuesday, November 04, 2008

More Autism Speaks epidemiology

Since last I wrote about Autism Speaks' epidemiology, two changes have been made on Autism Speaks' website. See here and here, compare to original versions here. No other changes were made, and Autism Speaks has recently elaborated on their epidemiology in press releases and the media. So it's possible to sum up Autism Speaks' epidemiology:

1. current prevalence of autism in the US is 1 in 150;

2. 10 years ago, prevalence was 1 in 1500 (as stated on all of Autism Speaks' press releases);

3. the "rate of autism is rising 10-17 percent annually. Unfortunately, the numbers appear to be continuing their upward climb";

4. there is an "epidemic" of autism, and this is a "recent epidemic";

5. because the autism epidemic is recent, older autistics do not exist; and

6. there are 1.5 million autistic people, total, in the US.

Given this array of information, the question remains as to how Autism Speaks calculated their 1.5 million figure. This figure is inconsistent with existing data. Autism Speaks' Chief Science Officer, Geraldine Dawson, was kind enough to provide more information:

1. Autism Speaks is using the 2000 census figure of ~280 million for current total US population;

2. Autism Speaks is using not the 1 in 150 prevalence figure Autism Speaks widely promotes, but a prevalence of 1 in 166;

3. the source for the 1 in 166 figure is Bertrand et al. (2001);

4. applying a prevalence of 1 in 166 across the total US population of 280 million yields 1.68 million autistics (it is actually 1.69 million);

5. this figure is rounded down to the nearest half million, to the Autism Speaks 1.5 million.

I can spot some problems. Autism Speaks states that the prevalence reported in Bertrand et al. (2001) is 1 in 166. But even a cursory reading of this paper's abstract shows that this is false. Here's a quote:

"The prevalence of all autism spectrum disorders combined was 6.7 cases per 1000 children."
Prevalence of 6.7 per 1000 comes out to 1 in 149--or about 1 in 150, the prevalence figure Autism Speaks widely promotes. Applying a prevalence of 1 in 150 across the 2000 US census figure of 280 million equals 1.87 million autistics. Following Autism Speaks' apparent policy of rounding to the nearest half million, that would be 2 million autistics in the US.

Then where does the 1 in 166 prevalence figure came from? The source Autism Speaks provided, Bertrand et al. (2001), does not report this figure. Autism Speaks' epidemiological expert, Michael Rosanoff, helpfully informed me that Autism Speaks was using figures from the US only, in calculating the Autism Speaks 1.5 million.

In fact the 1 in 166 figure first appeared in the literature in Chakrabarti and Fombonne (2001), as an estimate based on the results of three epidemiological studies. Two are UK studies (Baird et al., 2000; Chakrabarti & Fombonne, 2001); the other is the US study, Bertrand et al. (2001) which reported a prevalence of ~1 in 150. No published US epidemiological study reports a prevalence of 1 in 166.

Another problem is that Autism Speaks' Chief Science Officer is stating that there is a high, stable rate of autism. This is a scientifically sound position but, as enumerated above, one thoroughly rejected by Autism Speaks.

Instead, Autism Speaks widely disseminates the information that the prevalence of autism "has increased tenfold in the last decade." That is, 10 years ago, the prevalence of autism was 1 in 1500. If this is accurate, then the total number of autistics in the US would be less than 500,000. How much less would depend on what happened prior to 10 years ago, and how suddenly the leap from 1 in 1500 to 1 in 150 occurred.

At the same time, Autism Speaks is basing its epidemiology on three US studies. In Bertrand et al. (2001), the children were born between 1988 and 1995. Virtually all would be diagnosable as autistic by ten years ago, and the prevalence in this population was found to be ~1 in 150.

The more recently published CDC studies (ADDMN, 2007a, b) highlighted by Autism Speaks feature children born in 1992 and 1994. Again, virtually all would have been diagnosable as autistic by 10 years ago, and prevalence in this population was reported as ~1 in 150.

So Autism Speaks' array of information about autism prevalence is distant from what is reported in the scientific literature. Overall, Autism Speaks is claiming there are fewer autistics in the US than there actually are. How many fewer varies enormously, depending on which of Autism Speaks' contradictory figures is chosen.

And the upshot is primarily the denial of the lives and existence of older autistics. In an international "They Don't Exist" campaign, Autism Speaks is denying older autistics--around the world--recognition, rights, a voice, the kind of basic services nonautistics can take for granted, etc. This anti-scientific and unethical practice is a hallmark of autism advocacy. Autism Speaks is denying the existence of most autistic adults and of large numbers of autistic children.

In order to accomplish this amazing feat, this disappearing of autistics in the US and around the world, Autism Speaks is widely disseminating false, anti-scientific information about autistic people. This false information is deployed to raise money and to alter public policy according to Autism Speaks' goals. There is no thought as to the consequences for autistics. Autism Speaks is sending a powerful message that scientific findings--which Autism Speaks raises money to fund--should be dishonestly misrepresented and discarded, when these findings inconveniently get in the way of autism advocacy leaders such as themselves. And according to Autism Speaks, so should autistic lives be discarded, if they are in the way.


Autism and Developmental Disabilities Monitoring Network Surveillance Year 2000 Principal Investigators; Centers for Disease Control and Prevention. (2007a). Prevalence of autism spectrum disorders--autism and developmental disabilities monitoring network, six sites, United States, 2000. MMWR Surveillance Summaries, 56, 1-11.

Autism and Developmental Disabilities Monitoring Network Surveillance Year 2002 Principal Investigators; Centers for Disease Control and Prevention. (2007b). Prevalence of autism spectrum disorders--autism and developmental disabilities monitoring network, 14 sites, United States, 2002. MMWR Surveillance summaries, 56, 12-28.

Baird, G., Charman, T., Baron-Cohen, S., Cox, A., Swettenham, J., Wheelwright, S., & Drew, A. (2000). A screening instrument for autism at 18 months of age: a 6-year follow-up study. Journal of the American Academy of Child and Adolescent Psychiatry, 39, 694-702.

Bertrand, J., Mars, A., Boyle, C., Bove, F., Yeargin-Allsopp, M., & Decoufle P. (2001). Prevalence of autism in a United States population: the Brick Township, New Jersey, investigation. Pediatrics, 108, 1155-61.

Chakrabarti, S., & Fombonne, E. (2001). Pervasive developmental disorders in preschool children. JAMA, 285, 3093-9.

[Conflict of interest declaration: I'm affiliated with a research group which receives funding from Autism Speaks, among other funding sources.]


Anonymous said...

It is shameful for an organization to commit age discrimination. In doing this they position themselves along with the 'toxin' concern groups, another set of people that apparently wish to deny that autism conditions were documented by both Leo Kanner and Hans Asperger in the mid 1940s. By now many of their subjects would be over 60, some of them over 70.


Socrates said...

But what about the 1 in 100 figure, used by the UK government for planning:

Prevalence of disorders of the autism spectrum in a population cohort of children in South Thames: the Special Needs and Autism Project (SNAP) .

The Lancet , Volume 368 , Issue 9531 , Pages 210 - 215

G . Baird , E . Simonoff , A . Pickles , S . Chandler , T . Loucas , D . Meldrum , T . Charman

"Prevalence of autism and related ASDs is substantially greater than previously recognised. Whether the increase is due to better ascertainment, broadening diagnostic criteria, or increased incidence is unclear. Services in health, education, and social care will need to recognise the needs of children with some form of ASD, who constitute 1% of the child population."

Michelle Dawson said...

In response to Socrates, I wrote about Baird et al. (2006), which reports about the same cohort as in Baird et al. (2000), here, see the comments also.

A more recently published UK epi study, Williams et al. (2008), found a prevalence of about 1 in 161 (and a prevalence of 1 in 196 using a more rigorous standard of diagnosis). This is in a cohort born slightly later than the cohort in Baird et al. (2000, 2006), though the birth years overlap.

More information about Williams et al. (2008) is here.

Larry Arnold PhD FRSA said...

Well the prevalence is the proportion of whoever you decree to be autistic, for your knockabout compatriot Harold Doherty, it is one thing, for me it is another.

Autism seems to be one of the few conditions where the protagonists in the argument want to add an additional disability (that is medical model disability) threshold whereas that would be absurd elsewhere,

For instance diabetes is quite a serious impairment of funtion, however controlled diabetes is not as much of a disability as it would be without insulin.

The facts are however that true prevalence can never be established simply because it is such an imprecise phenomenon that the epidemiologists are attempting to count, it is like trying to establish the prevalence of Englishness from whether one knows the laws of cricket or not, it doesn't work.

Autism is too ill defined a multi dimensional phenomenon to be counted, basically because of the way it has been socially constructed and due to the inevitable bias of the observers.

Terry Brugha thinks he can get over this, good luck to him, but to me he is another ignorant shrink who thinks he knows better than the entire academic consensus of sociology, and of disability studies which will tell you that there exists a sort of Heisenberg effect, where the observation of the phenomenon itself effects what it is one is observing.

It is all about perception and the social need to construct a view of autism which is essentially intertextual that is to say which cannot stand on its own and is dependant upon so much else. In the end it is not much more than bricolage.

Michelle Dawson said...

Yes, I'm already aware that only Mr Rex knows the truth. Mr Rex says so quite often. I bow down to his vast superiority (Heisenberg? of course!) and perfection.

But I'm afraid I'm going to keep plodding along in my terribly inferior manner, reporting accurately from primary sources, and taking the risk of testing ideas in the tremendously imperfect way (but less imperfect than all the other ways) called science, and working towards more (rather than less) accurate information, and better (rather than worse) decisions.

Socrates said...
This comment has been removed by the author.
Socrates said...

Thanks for those links MD.

However, my ignorance, and cynicism after more than 30 years experience of the psychiatric system, gives me little faith in any of these people being able to come up with a truly meaningful figure.

For example, only today I read in Pulse[ http://www.pulsetoday.co.uk/story.asp?sectioncode=18&storycode=4121057&c=1 ],

a piece co-authored by Baron-Cohen aimed at General Practitioners which includes this phrase:

"Not everyone on the autism spectrum needs a diagnosis – this should be reserved for those who need it to access help"

Let's hope Brugha and his merry band of epidemiologists don't feel that not everyone on the spectrum needs to be enumerated..

Anonymous said...

Larry is correct, nevertheless. If he weren't, how do we explain the prevalence of 4.5 in 10,000 determined by Vic Lotter's screening in 1967? An epidemic?

Different criteria, I'm sure. But criteria is the backbone of the social construction of autism, clearly. This confirms Larry's observation, rather than contradict it. The evolution of the construct over time is undeniable.

It's not only criteria that drives it too. You suggested this yourself above, Michelle. Why do Baird et al. and Williams et al. get divergent results in the UK? Different researchers, different methodology, different ways of interpreting what autism is.

Would it be reasonable to take 1/3rd of Baird et al.'s autistics and tell them, well, you don't meet threshold according to this other methodology; you're not "really autistic" ?

Michelle Dawson said...

While science may be plodding and messy and all (compared to grandly declaring the truth), it is still the best way to find things out.

E.g., you can read the papers and look at the methodology and the data. Also, you can read the accumulated body of work in an area (like epidemiology), an area in which research of various quality will continue to accumulate.

Given that different papers have different methodology, you would not expect them necessarily to have identical findings. And you might also find evidence that small changes in different aspects of methodology have a major effect on reported prevalence, across studies.

This knowledge allows for better interpretation of studies, something I suggest is better than discarding an entire area of research (or all research) because scientists and scientific papers are flawed. As they all are, of course.

Mr Rex opposes this difficult, messy, incremental process, and wants to replace it with a declaration of the pure truth according to those claiming to embody it. I disagree with this approach.

I don't know which Baird et al. paper Joseph is referring to (there is more than one). But in my view, it is possible to work towards more accurate, rather than less accurate diagnosis, and better quality, rather than poorer quality, epidemiology. This will never meet Mr Rex's standards of perfection, but this does not mean the effort should not be made.

Socrates said...


as an ignoramus, I need to start from basics.

The question I need first to answer is "What is Autism".

I'm afraid the DSM IV 299.8 doesn't work for me...

Michelle Dawson said...

In response to Socrates, see this post.

I wrote that a long time ago, but it's not too bad.

andrea said...

Apparently the wags at Autism Speaks don't understand the difference between prevalence and diagnostic rates!

And great point, Patrick, about the 1940s autistics!


Socrates said...

Once again thanks for the link.

I was astonished to see a comment I left on the doctors site Pulse, re: the above mentioned article by Baron-Cohen was approved.

* Harry Williams | 04 Nov 08

I question the wisdom of this point:

"Not everyone on the autism spectrum needs a diagnosis this should be reserved for those who need it to access help"

Life crises are often grossly aggravated by the presence of autistic spectrum conditions (which may present initially in a 'sub-clinical' form), and the absence of such a diagnosis, where justified in a more academic context, may leave health practitioners without a valuable piece of information.

Larry Arnold PhD FRSA said...

Michelle you really do yourself no favours in assuming there is one monoglot monolithic supreme method that is science which will draw back the curtain from all darkness.

In fact there are a number of somewhat overlapping scientific epistemologies each of them having evolved to suit a particular kind of research, and I am afraid that Autism research lies somewhat closer to the "wooly" epistemologies of social psychology than you would like to think.

Even if you were to subsume (horror of horrors) autism into a branch of medicine, you would find that medical research is somewhat less certain than physics or astronomy.

Experimental method, well there is more than one...

Statistical analysis, I think we can all see, and do see the tricks that are played with statistics, and the reason for that is that numbers are inherently manipulable.

They behave according to certain rules, but then they are made to behave that way, formulas are created to approximate, analyses of variance, well there is no one method, you use whatever best suits.

In any case I wonder what stats would look like if you used harmonic means instead of artithmetal ones, quite different I am sure.

What seems to have as big an influence on research as who one includes in ones criteria, is as often as not who one arbitrarily decides not to include. How you draw the lines does affect the data, and if you pit autism study against autism study, you are not dealing with the physical certainties of real hard science, where a chemical reaction can be measured, or a definate equation adduced for the amount of light falling on an object at a given level of illumination and distance, or for the point at which a given material will fail under tension or compression.

No Autism "science" has neither the precision of optics or engineering.

Michelle Dawson said...

Mr Rex wrote:

"one monoglot monolithic supreme method that is science which will draw back the curtain from all darkness."

My question is, where have I written such a grand thing? I can barely read it.

Larry Arnold PhD FRSA said...

You do so with every instance where you use the word science, as if science were one thing, and one thing only, like gnosis.

Michelle Dawson said...

Science is a methodology. That's all it is.

Philosophers have attached all kinds of fancy things to science, as they have to ethics. But these attached fancy things should not be mistaken for science or ethics.

I have no clue as to the many other grand, intellectual, post-modernist (meaning, "we are way too good for science" or words to that effect), etc., things Mr Rex is going on about.

Anonymous said...

For the German translation please visit http://autismus-kultur.de/autismus/politik/epidemiologie-von-autism-speaks.html.