Monday, April 27, 2009

Dr. Jay Gordon Caught Making Stuff Up?

Just when I thought I wouldn't encounter better examples of the intellectual bankrupcy of anti-vaxers, comes Dr. Jay Gordon, pediatrician to Jenny McCarthy's son, with the following anecdote:
I just saw an eight-month-old boy who got two vaccines then lost his language, motor skills, reaction to his own name and responsiveness to his 3 year old sister. This all happened three hours after a DPT/HIB combination. All previous observation and video of this little boy is normal. Lots of great video of a vibrant, talkative happy baby.

He has autism now. No proof, but the temporal proximity of the regression to his vaccines is daunting to those of us who would like more rigorous proof rather than just a collection of hundreds of pieces of anecdotal evidence. Tempting to assume causation even if only this one case/child.

Just thought you'd want to know. I usually see these kids in my office months or years after the event the parents think caused the problems. I have read hundreds of emails and spoken to thousands of parents who are certain that vaccines triggered or mightily contributed to their children's autism.

Not this time: This family came to my office just days after the vaccines, distraught that he was no longer talking, smiling, acknowledging his sister or his mom and dad and was flapping his hands a lot.

(Respectful Insolence - Comment)

Don't worry. I waited for confirmation from Orac before writing this post. Some people doubted that was actually Dr. Jay, but it is. Since he posted at 00:44 Pacific, there's some speculation he might have been drunk, but who knows.

That is monumentally absurd, isn't it?

First of all, who in the hell diagnosed an 8 month old with autism? That simply can't be done. Even an 18 month diagnosis is pushing the envelope.

Then we're supposed to believe an 8 month old "lost his language" and that this was noticed in a matter of 3 hours. Typically, 8 month old babies do not talk. This is completely normal.

In so far as responding to his own name, this is possible for an 8 month old. The "name test" is supposed to be done on the baby's first birthday, however (source).

Eight month old babies have little in the way of motor skills that they could noticeably lose, unless the baby becomes paralyzed or something like that. A baby of that age can't yet walk. He should be sitting by then. In any case, loss of motor skills is unrelated to autism, and in fact should be cause to consider explanations other than autism. If it was a really obvious loss of motor skills, I'd think the baby should be rushed to a hospital, but instead the parents decided to see Dr. Jay again in 3 days.

Enjoyment of play with other children should occur around the age of 2 years.

Hand flapping is entirely normal in infants, and it seems to me that, as a pediatrician, Dr. Jay should be aware of this. Plus, in order to hand-flap, the baby needs some motor skills.

The DTP vaccine is no longer used in the US. The DTaP vaccine is what is used. It's administered at 2, 4 and 6 months of age. Same with the Hib vaccine.

So what do we have here? Did Dr. Jay make it up completely? Perhaps this is a non-event that was blown out of proportion.

Now, I wonder, was this child vaccinated by Dr. Jay? If not, why would parents take a child with the apparent symptoms noted to a doctor who is not qualified to diagnose developmental disorders, in just 3 days? Then again, is it plausible that Dr. Jay would administer 2 vaccines in one visit?

I suppose that a pediatrician who scares parents about autism and discourages vaccination could cause them to notice things that are completely normal all of the sudden.

Saturday, April 25, 2009

Maxine Aston and CADD - A Different Level of Crankery

I'm sure readers are aware fellow blogger Alyric passed away last week. Subsequent to her passing, there has been some discussion about something called Cassandra Affective Deprivation Disorder (CADD), which Alyric had felt strongly about. (See the ASAN Petition, this post by Gonzo, and this older post by ABFH with over 60 comments.)

I have to admit this CADD thing had completely passed me by. It was only after I read the documentation related to the recent activity that I realized it pertained to yours truly in a personal sense. But I wanted to see if the claims around CADD had any merit from a more objective point of view, personal implications aside.

The Syndrome

Maxine Aston claims to have discovered a new syndrome called CADD, observed in the significant others of autistic adults. Unfortunately, the only definitions of the syndrome are found in websites and books that simply list the characteristics of the alleged syndrome. I could find no mention of it in the scientific literature.

I haven't found a "how to" on making up new syndromes. But it's possible to research how it has been done historically.

In order to define a syndrome, it's not sufficient to just come up with a list of symptoms. You have to show that it exists. Testimonials are not sufficient. Consider Kanner (1943), which defined autism for the first time. Kanner wrote the case histories of 11 autistic children in detail, based on meticulous research observations.

This was not sufficient, though. For "autism" to be considered valid, subsequent research was required. For example, if autism had been found to be the same as other constructs like mental retardation or schizophrenia, say, when you match groups by IQ, then it would've been deemed an invalid construct.

The syndrome should also have some usefulness (e.g. Does it help predict something useful? Does it help define unique needs the individual might have?)

In fact, today there's some controversy about the validity of Asperger's Syndrome, since the differences between HFA and Asperger adolescents and adults (when matched by IQ) are nearly non-existent. Asperger's could very well be an unnecessary construct. But I digress.

For CADD to be considered valid, there not only needs to be a write-up of case histories in the scientific literature, you also need to show that it is different to everything that has been reported previously. Unhappy significant others and failed relationships abound. Are the relationships with autistic adults unique in this sense, say, when matched for socio-economic status and disability in one partner? And if so (which I don't doubt is true) are they uniquely bad, or are they simply different? Can any problems be explained by lack of acceptance?


The existence of the syndrome has not been established, so it seems a little spurious to talk about causes. Nevertheless, let's look into what the alleged causes are.

As far as I can tell, "sufferers" of CADD claim that it is caused by lack of empathy and lack of emotional reciprocity in the part of the autistic significant other.

Regarding empathy, there's some recent research that shows adults with Asperger's Syndrome score lower in measures of cognitive empathy, but not in affective empathy. Persons with Asperger's actually experience higher levels of distress when observing others in distress.

These findings bring into question the means by which empathy is measured in research.

I'd also recommend reading Dr. Gernsbacher's Toward a Behavior of Reciprocity.


The symptoms of CADD are claimed to be essentially the same as the symptoms of Seasonal Affective Disorder (SAD), including sleeping problems, depression, etc.

Again, in order to make these sorts of claims, some evidence, perhaps in the form of a case-control study, is needed. Testimonials are not sufficient. It would also be necessary to show that the symptoms can't be explained by socio-economic status, assortative mating, or disability in the significant other.

There are more far-fetched claims attributed to Aston, like CADD resulting in cancer. I'm sure this is based on one or a few anecdotes. Since cancer is fairly common in adults (the lifetime risk of developing some form of cancer is over 40%) such anecdotes are nearly useless.

Besides, there's some evidence that psychological stress does not contribute to cancer risk (e.g. Johansen & Olsen, 1997). The matter appears to be unresolved.


What we have here is a syndrome that has not been shown to exist, with alleged causes that might be erroneously attributed to autism, and with symptoms that haven't been shown to be associated with the alleged causes.

Sound familiar? This is no different to crankery like "autistic enterocolitis" or "mercury-induced autism." Except the promotion of CADD has the potential to result in a lot more direct damage to the personal lives of autistic adults.

Monday, April 20, 2009

Adult Outcomes from Toronto and Utah. Then and Now.

After some recent discussion around the performance of autistics in college, I decided to read Szatmari et al. (1989). I was aware this adult outcome study looked at the academic achievements of autistics without intellectual impairment (IQ >= 70).

I found that this 20 year old paper has a lot of other information of interest. Additionally, the population it studied is similar to that of the recent outcome study out of Utah that Sullivan had written about over at LB/RB.

Even though the Utah paper is apparently not yet available, there is some information from it which we can use to compare and contrast. I thought it would be useful to go through Szatmari et al. and see how it compares to the Utah study when possible, and to other outcome studies I'm familiar with.

Group Sizes

One limitation of Szatmari et al. is that it was small. They only managed to find 16 autistics who agreed to participate. The Utah study included 41 adults.

General Outcome Assessments

Szatmari et al. doesn't really provide ratings of "good" or "very good" but a fair assumption is that half of the adults would've been given these ratings, which is basically the same thing the Utah study finds.

It also reports that 4 of the 16 (25%) might be considered "recovered." This is how the paper puts it:
If a naive individual were to meet the first four probands from Table II without knowledge of their early history, one might conclude they were essentially normal.

The Utah study reports that 6 of the 41 (15%) no longer had autism diagnoses. As Sullivan notes, we don't know if they have Asperger or PDD-NOS diagnoses.

Employment & Independence

Szatmari et al. finds that 8 (50%) were in full-time employment, 3 involving regular contact with people. Five were living independently, 10 with parents, and 1 in a group home.

In the Utah study, half had full-time or part-time employment. About half lived in group homes or with parents.

Dating & Marriage

Szatmari et al. reports that 4 individuals had dated regularly and had been in long-term relationships. One was married.

The media report of the Utah study states: "A few are married and have children. They have friends or acquaintances."

Predictors of Outcome

The usual predictors of outcome, like IQ, are noted in Szatmari et al. What I wanted to go over are some of the predictors and non-predictors, measured or qualitative, which are not well known but have been mentioned elsewhere.

On early severity as a predictor, Szatmari et al. says:
Nevertheless, the results do suggest that, for this group of HFA probands, early history explained little of the variance in outcome. Indeed, the good and poor outcome groups differed little with respect to early impairments in social responsiveness, deviant language, and bizarre behaviors.

From the Utah study:
While all participants had baseline IQs in the nonimpaired range, there was limited evidence to support the use of other early childhood variables to predict adult outcome.

About the time frame when most gains were observed, Szatmari et al. tells us that:
Parents noted that the good-outcome cases had improved considerably by late adolescence. By this age, parents felt their children were still somewhat shy and awkward but not odd or eccentric.

This is something I had previously read about in Kanner (1972):
They have not completely shed the fundamental personality structure of early infantile autism but, with increasing self-assessment in their middle to late teens, they expended considerable effort to fit themselves — dutifully, as it were — to what they came to perceive as commonly expected obligations.

It was not until the early to middle teens when a remarkable change took place. Unlike most other autistic children, they became uneasily aware of their peculiarities and began to make a conscious effort to do something about them.

Szatmari et al. has something to say about the involvement of the parents:
The mothers of these probands were, however, able to advocate forcefully for their child in terms of educational and recreational resources.

A similar observation can be found in Ruble & Dalrymple (1996):
What seemed to be an important predictor of success was that whenever individuals and their families were confronted with challenges, they sought and successfully accessed various supports. For example, when families were told to "place" their young children, their parents sought and created alternatives. Some of the families were the first to push for integration in school and used their natural community and family ties to include and support their child.

College Outcome

Unfortunately, we don't have data on college attendance from the Utah study. I will go over what Szatmari et al. finds, and also some data from Kanner (1971) and Kanner (1972).

Eight (50%) of the group from Szatmari et al. had attended college or university. Seven had obtained degrees. The remaining one was beginning his second year as an undergraduate.

I find this quite interesting, statistically. As of 1990, about 42.7% of persons 18-19 in the US attended college (source). It was 38.5% for 20-21 year olds. Of course, when it comes to Szatmari et al. we're talking about Canada, not the US. Additionally, the autistic individuals came from affluent families, according to the paper. Diagnosed autistics tend to be urbanites.

Now, as of 2004, about 33% of college students in the US graduate in 4 years or less, 50% in 5 years or less, and 55% in 6 years or less. About 45% take 7 years or more to graduate, or never do (source.) I think it's remarkable that none of the autistics from Szatmari et al. who attended college had failed school at the time of the study. Only one (age 19) had not obtained a degree yet, but was still in school.

Note that the autistics from Szatmari et al. were not a bunch of geniuses by any means. The average IQ of those autistics who attended college was apparently 102 (range 86 - 110.) This is similar, probably even less than the IQ of the average college student.

They generally did not obtain graduate degrees. One had an MBA. A couple had Bachelor of Science degrees. This is probably not surprising statistically. Plus, again, the study is 20 years old.

College was a relatively adequate environment for autistics, according to Szatmari et al.
Living in residence at university seemed to make a major difference to the probands themselves. It was as if the forced confinement of living with others had a beneficial effect on their socialization. Certainly, this was a time when the probands felt most accepted by their peers. To many, these were their happiest years.

I also scanned Kanner (1971) and Kanner (1972) for mentions of autistics who had attended college. I found 7 such persons, not counting "the gifted student of mathematics killed accidentally and the young man whom we have so far lost track after 1962 when he was in college." Of the 7, 5 had obtained some sort of college or university diploma. One of them had a Master's degree in Economics. One was said to be struggling in general college, but there's no follow-up as to outcome. One was a student, doing well at university, and gifted in Math.

One issue with the Kanner papers is that they were qualitative in their entirety, so it's possible some of the autistics he saw went to college and failed, but he didn't write about that.


When you look at the outcomes of autistics who have a distribution of IQs similar to what you find in the general population, certain difficulties and impairments will be evident: autistics struggle a lot when it comes to marriage, dating, friendships, independence, etc.

When it comes to employment, we know that autistics are under-employed. It's not clear how well the autistics who are employed do at their jobs. Some qualitative data from Kanner (1972) suggests that autistics who grew up in the 1960s and who got jobs did fairly well. It's also not clear why half the autistics from Szatmari et al. were not employed.

As far as college enrollment, there doesn't appear to be much of a difference between autistics and non-autistics of the same intelligence. In fact, autistics seem to do (or used to do) rather well in college, judging by the rates of graduation reported.

Tuesday, April 07, 2009

Very Quick Note About Self-Concept and Achievement

I was looking for something else entirely and I happened to come across an old paper titled "Self-Concept of Ability and School Achievement" (Brookover & Thomas, 1964.) I thought it was pertinent to the discussion around the BRAINHE study on neurodiversity and particularly Jake Crosby's contention that self-confidence does not have an effect on "real" success. His claim was counter-intuitive all in itself. This might also be of interest to Billy Cresp, who appears to believe intelligence is the only variable that matters.

The abstract follows.

Three hypotheses concerning self-concept were tested using a sample of 1,050 seventh grade students and a selected subsample of 110 over- and under-achieving students. A significant positive relationship was found between self-concept of ability and grade point average; this relationship persisted even when measured intelligence was controlled. Specific self-concepts of ability related to specific areas of academic achievement were found; in some areas these were better predictors of achievement in the subject than general self-concept of ability. Self-concept was significantly and positively related to the perceived evaluation of significant others. A direction for further research is indicated.

(Emphasis mine)

Monday, April 06, 2009

One Handley Turd I Left Out

In my previous post I looked at a few statements JB Handley made during his recent appearance on Larry King Live. Whether the statements were dishonest or simply wrong, I can't tell, but they were notable in their wrongness. They have even been referred to as fractally wrong.

I left an important one out, which I was pretty sure was wrong, but I needed to research it a bit more.
So I would tell you either go back to the 1989 schedule before this whole mess happened or go to Finland's schedule, Sweden's schedule, Norway's schedule and use theirs. Because their autism rates are one tenth of ours.

He must mean their rates of identified autism are one tenth of what they are in the US. I don't like it when people don't make that clarification.

But let's see what a quick search of the literature has to say.


One study in the town of Karlstad, Sweden (Kadesjö et al. 1999) had found the prevalence among 7 year olds to be 1.21%. I'll let Lorna Wing summarize why this study is of interest:
Kadesjö et al (1999) report a study in Karlstad, a Swedish town. Although this was small scale it was very intensive (over 50% of the 7 year old children seen and assessed personally by the first author). The study found a prevalence for all autistic spectrum disorders for all levels of IQ, of 1.21%!!! Children were followed up four years later and had the diagnoses confirmed.

That's not all, though. You have, for example, Arvidsson et al. (1997), a study that set out to identify "all individuals with severe degrees of autistic behaviour." They found a prevalence of 31 in 10,000 among 3 to 6 year olds. This is more like half of the consensus prevalence in the US.

Then there is Gillberg et al. (2006) which finds a prevalence of 20.9 in 10,000 for autistic disorder, and 32.9 in 10,000 for "other ASDs." That's 53.8 in 10,000 total.

There are several other studies from Sweden we could look at, but I think the wrongness of Handley's claim has been clearly established just with the ones mentioned.


Mattila et al. (2007) found a prevalence of 25 in 10,000 for Asperger's Syndrome alone. This is actually higher than what the prevalence of Asperger's in children is normally found to be.

An older study, Kielinen et al. (2000), finds a prevalence of 20.7 in 10,000 among 5 to 7 year olds. This is lower than the US consensus prevalence, but it's only 1/3rd of it, not 1/10th. Additionally, note that 50% of the children had IQs below 70. This is different to what you see in California DDS, for example.


The most recent epidemiology from Norway is a bit old, and here we do see that the prevalence is similar to what it traditionally was for Kanner autism. For example, Sponheim & Skjeldal (1998) find a prevalence of 4 to 5 in 10,000 for 3 to 14 year olds, using ICD-10 criteria.

However, there's a recent screening of 7 to 9 year old children using the ASSQ. That's Posserud et al. (2006). It finds that 2.1% of children were high scorers in the ASSQ when both teacher and parent questionnaires were considered. (It was 2.7% if they considered teacher forms only, as some parents declined to participate.) That seems high.

Unfortunately, Posserud et al. don't tell us how many were confirmed to be diagnosable with ASD after the screening. There's an update of that study, Posserud et al. (2009). I wish I had a copy. All I know is that the ASSQ is found to have a sensitivity of 0.91 and specificity of 0.86, which appears good.

In any case, JB Handley's assumption is kind of simplistic even if we only consider the country of Norway, as it ignores all the issues involved in identifying autism.

Saturday, April 04, 2009

JB Handley - Poorly Informed or Outright Liar?

Last night I was changing the channel and I stumbled upon Larry King Live (transcript here.) It must have been only a couple of minutes before they switched to Anderson Cooper. During that time, I got to watch JB Handley make a monumentally false statement. I'll go over several statements he made during the show, but first I want to address the one that caught my attention.
Did you see the January study from UC Davis using California's numbers that said unequivocally there's been a clear rise, it's not do [sic] to diagnostic substitution.

Handley is referring to Hertz-Picciotto & Delwiche (2009). This is what the conclusions of the paper actually said:
Autism incidence in California shows no sign yet of plateauing. Younger ages at diagnosis, differential migration, changes in diagnostic criteria, and inclusion of milder cases do not fully explain the observed increases. Other artifacts have yet to be quantified, and as a result, the extent to which the continued rise represents a true increase in the occurrence of autism remains unclear.

That's not an unequivocal determination of a clear rise, by any stretch of the imagination. Besides, that paper was surprisingly poor. None of the figures used by the paper approach anything that might be called certain or accurate. One of the artifacts the paper fails to take into account is probably the most pertinent artifact of all: awareness. A summary of all the very serious problems the paper had can be found at the end of this post.

Next statement:
We're looking for something that's caused this epidemic. It went from one in 10,000 in the 1970s to less than one in 100 today in many states.

The first epidemiological study of autism was Lotter (1966). It finds a prevalence of 4.5 in 10,000. That was in the UK. Wing et al. (1976) finds the same prevalence in the US. Mike Stanton has a good summary of historical prevalence studies. I don't think this is the first time I've corrected Handley and others about this.

I happen to have read Lotter (1966). What they called 'autism' back then is very different to, say, what the DSM-IV calls 'autism' today. It's apples and oranges. Here's a excerpt from Lotter (1966) to give you an idea:
The two categories of the Creak criteria concerning a "'pathological preoccupation with particular objects" and "an insistence on the preservation of sameness" were therefore combined.

These are similar to criteria C2 and C4 of the DSM-IV, but in the DSM-IV they are basically optional (only 1 item from C is required.)

The paper explains they located 666 out of 75,930 (88 in 10,000) children who had "certain kinds of behaviour" that couldn't be simply characterized as "backwardness." They initially excluded 87% from this group, based on the opinions of two judges familiar with the syndrome of autism.

They ended up with 32 autistic children, 22 (68.8%) of whom had IQs under 55, another 5 (15.6%) with IQs between 55 and 79, and the remaining 5 (15.6%) with an IQ of 80 or above. Again, this is quite different to what we call 'autism' today.

Moving on:

I want to talk about this issue of autism prevalence. It's going to be shocking for parents to learn that the CDC and the AAP don't actually acknowledge that there's been a real rise in autism cases. Larry, the Department of Education in 1992, 16,000 kids were getting autism services. Today 225,000. That means in 1992, they were missing 93 percent of kids with autism. Where are all the adults with autism? They don't exist.

Adults with autism don't exist? That's not only insulting and potentially damaging to the quality of life of autistic adults; it's completely wrong. It becomes embarrassingly wrong every time someone goes and surveys adults to see if autism can be found among them.

Again, I doubt this is the first time this has been pointed out to JB Handley.

Heck, Handley's Age of Autism has a token autistic adult who contributes to that blog from time to time.

His Special Education figures are essentially correct, except the year is 1993, not 1992. What he fails to mention is that 489,000 students were receiving services under the mental retardation category in 1993, and only 425,000 as of 2007.

Let's consider Specific Learning Disability and MR together, and let's also consider population growth. The prevalence of both categories combined was 645 in 10,000 in 1993, and 584 in 10,000 as of 2007 for 6 to 17 year olds. That's a drop of 61 in 10,000 – more than enough to cover the increase in the administrative prevalence of autism.

Is it possible that JB Handley has never heard of diagnostic substitution in IDEA? Doubtful.

Those are all the statements I will address in this post. Readers are welcome to point out other errors, if any.

My impression is that JB Handley is either completely misinformed about some basic facts having to do with autism epidemiology, or he decided to go on Larry King to simply lie with a straight face.

To conclude, here's some advice for Larry King: In the future, if you want to discuss the "autism epidemic," you would do well to invite recognized experts on the matter, such as Roy Richard Grinker or Eric Fombonne.

Friday, April 03, 2009

Autism Awareness and its Relationship to the "Epidemic"

Yesterday was World Autism Awareness Day. (That's right, its acronym is WAAD, and it's the day after April Fool's day.) Given the occasion, I thought I would discuss "awareness." More specifically, I wanted to discuss how we can tell that awareness has had an impact in what is usually referred to as the "autism epidemic."

That awareness could account for much of the increase in autism diagnoses since the early 1990s is taken for granted. You will find awareness mentioned in many autism papers. In the recent MIND Institute paper (Hertz-Picciotto et al. 2008) awareness was acknowledged but not taken into account in the calculations. While H-P et al. apparently did not believe awareness to be an important factor, many other researchers seem to think it is. I personally believe it's supremely important. Changes in criteria can't in themselves cause the number of diagnoses to rise. Autistic persons need to be found and diagnosed (with autism) before they are counted. If diagnostic substitution is increasingly occurring, there's probably a reason why it's occurring: awareness.

Nevertheless, as far as I know, no one has ever demonstrated that awareness explains the rise, statistically. Thus, we fall into the same trap that proponents of most environmental hypotheses fall into. We assume that because awareness has increased in the information age (as it obviously must have) and rates of autism have also increased, one causes the other. Could it simply be a coincidence?

How do you measure awareness? Here's one way. I will use Google News Archive Search to search for the phrase "autism california" every year from 1992 to 2006. I propose that the number of results is an adequate proxy of autism awareness in California, though like all proxies, it's just an approximation.

I'd like to have data on diagnoses of autism to be able to make a comparison. I don't have that, however. Here's the next best thing. I will use data on number of California DDS clients with a classification of autism as reported on March of every year from 1993 to 2007. More specifically, I selected the number of autistic children born only 3 years prior to the report.

Let me explain the rationale. If we were to choose 5 year olds, that wouldn't work very well, as they could've been diagnosed when they were 2, 3, 4 or 5. It would be difficult to correlate caseload in the report year with awareness in several prior years. If we instead choose children born only 3 years before the report year, they almost necessarily have to have been diagnosed the year prior to the report year. This is especially true if we use the March report. If we were to use the December report, this again might not work very well.

Without further ado, the following is the graph I came up with.

The blue line represents awareness. The green line represents autism caseload of 3 year olds, approximately. There are 2 Y axis scales, left and right respectively. I think the 1 year lag between one and the other is pretty clear. If you prefer to visualize the data in a scatter chart, here it is:

It's true that correlations that are purely coincidental (e.g. pirates vs. global warming) can result in convincing scatter charts like the one above, and I haven't done further statistical analysis to rule out coincidence, but I think there are a couple of clear features of the series that make me think we're looking at causation. First, I already noted there's a 1-year lag in the expected direction. Second, awareness seems to start to shoot up in 1999. Autism caseload, in this cohort of very young children, starts to increase noticeably in 2000. Then there's also a clear spike in awareness in 2005, which corresponds to a spike in caseload in 2006.

Considering there's also plausibility, I think we can safely say awareness did have an impact in the rise of autism diagnoses. I only looked at very early diagnoses, but this is because I don't think I have other data that could be adequately used instead. I would imagine awareness has an impact on diagnoses at any age.