I have just started watching the video of the debate between David Kirby and Arthur Allen on the subject of autism and vaccines. Kirby wrote a book, Evidence of Harm: Mercury in Vaccines and the Autism Epidemic: A Medical Controversy which starts from the premise that the rapid growth in recorded cases of autism in the USA that began in the early 1990s represented a real growth in numbers and could not be explaind by increased awareness, better diagnostic procedures or a change in the criteria. There had to be an environmental trigger. At the same time changes in the mandatory vaccination programme for children in the USA increased exposure to thimerosal, a preservative that contains ethyl mercury.
Did the thimerosal cause an autism epidemic? Kirby believes it did. Allen once thought it was a credible hypothesis. But in the course of researching his book, Vaccine, the Controversial Story of Medicine’s Greatest Lifesaver, Allen changed his mind. Allen has also begun to question whether there has been an epidemic or not. See his review of Roy Grinker’s book, Unstrange Minds, Remapping the World of Autism.
I was looking forward to the debate. But within minutes I was hitting the pause button and diving into my archive of autism related papers. Kirby began, quite rightly, with a discussion of epidemiology. But his version seemed at odds with what I thought I knew. Still, this was an important debate. Kirby must have checked his facts beforehand. So I went to check mine.
Kirby’s Fact 1.
In the 1980s autism prevalence in the USA was between 1 and 2 in 10,000
If anyone has a reference to an epidemiological study for this extremely low figure I would love to have it. When Lorna Wing surveyed the major epidemiological studies carried out between 1966 and 1992 she referred to two studies in Utah (Ritvo et al 1989) and North Dakota (Burd et al 1987) that found rates of 4 and 3.3 in 10,000 respectively for DSM III autism which use very similar criteria to Kanner’s criteria. She also mentioned a study by Treffert which found a prevalence of 3 in 10,000 in 1970 in Wisconsin. When I considered Wing’s survey in an earlier post I remarked upon the robust nature of the figures. Researchers who combined consistent epidemiological methods with Kanner’s diagnostic criteria found rates of between 4 and 5 in 10,000.
Kirby’s Fact 2.
In the late 1990s autism prevalence in the USA was 20 in 10,000
Kirby offers no citations for this figure. Probably the most well founded study in the USA in the 1990s was by Yeargin-Allsopp et al in Atlanta in 1996 which found a rate of 34 in 10,000 in 3-10 year olds. It was reprinted in JAMA in 2003. In the same edition Fombonne considers this an underestimate and thinks the 5 to 8 year olds in the study provide a more reliable estimate of 41 – 45 in 10,000. (JAMA 2003 Volume 289 Issue 1.)For comparison Wing and Gould found a rate of 20 in 10,000 in the Camberwell study in 1979. This study confined itself to children with learning dificulties in special schools and never looked at children in mainstream classes where most autistic children are found today.
Kirby’s Fact 3.
By 2000 autism prevalence in the USA was 40 in 10,000
Again there is no citation. And autism prevalence in whom? What is the age cohort?
Kirby’s Fact 4.
By 2004 autism prevalence in the USA was 60 in 10,000
We do have a lot of epidemiology for this figure.It is supported by the Medical Research Council in the UK and the Centers for Disease Control in the USA. But it is usually taken to mean that we have now reached a reasonably accurate estimate of prevalence figures for the entire autistic spectrum across the entire population. Kirby sems to be using these figures to suggest a year on year increase in incidence which is not the same as prevalence. Generation Rescue made a similar mistake last year which I commented on.
When you are dealing with statistics it is important not to get incidence and prevalence muddled up. Incidence refers to the number of new cases in a population over a period of time. Prevalence refers to the total number of cases in a given population at a specific time.
Kirby’s Fact 5.
By 2004 th USA figure of 60 in 10,000 was the same as in the UK. But the USA had 40 in 10,000 with autstic disorder and only 20 in 10,000 with other ASDs. In Britain the figures are reversed: 20 in 10,000 with Autistic disorder and 40 in 10,000 with other ASDs.
Kirby’s explanation is simple. Here in the UK we only ever had half the thimerosal in our vaccines compared to our American cousins. So we only got half the autistic disorder. OK. Thimerosal causes autistic disorder. There is a linear relationship. Double the thimerosal and you double the autistic disorder. So what causes PDD-NOS and Asperger Syndrome? Why should they be twice as prevalent in the UK as in the USA? Is that environmental or genetic? Perhaps we Brits are naturally more high functioning than the Yanks 🙂
By this time I was beginning to get a little bit sceptical about Kirby’s figures. He actually did mention a source for his 40 in 10,000 with autistic disorder, Brick Township. I remember that one; 40 in 10,000 for autistic disorder is correct. But the figure in Brick Township for other ASDs was not 20 in 10,000. it was 27 in 10,000. 67 in 10,000! That is a lot of autism, except for one fact. It was a very small sample; 60 children aged 3 to 10. The authors acknowledge the problems in generalizing from their data.
As mentioned earlier, the major limitation of this study was an inability to ascertain higher functioning individuals who were not in any special education class in public schools or had not been seen by participating clinicians. Consequently, because of these case-finding limitations, the results from Brick Township must be considered a minimal prevalence for autism. Categorical distinctions between autistic disorder and the other ASD were probably limited because the ADOS-G has been found to over estimate autistic disorder relative to PDD-NOS. Also, because clinical assessments could not be conducted for 17 children and the diagnosis had to be based on records alone, the reliability and validity of the diagnosis for those children is limited. Discrimination between PDD-NOS and autistic disorder also may have been influenced for these cases given that over 56% of the children who participated in the clinical assessment were determined to have autistic disorder in comparison to only 27% of the children assessed by record review only. Finally, the prevalence rates for autism obtained in this study must be generalized with caution since the community was selected for study because of a suspicion of increased numbers of children with the disorder. Studies of larger populations, such as one that included surrounding communities, may yield different findings.
They may indeed.
Kirby’s Fact 6
Denmark removed all its thimerosal way back in 1992 and it has a rate of only 8 in 10,000.
Yes, except that according to this study autism rates went up in Denmark after they removed the thimerosal. Actually the base rates for autism in this study were so low it is ridiculous, less than 1 in a 1000 throughout the seventies and eighties. That is less than Brask found in 1972 in Denmark for Kanner’s autism. (4.3 in 10,000) A more recent study suggests the real rate for ASDs in Denmark is closer to the 34 in 10, 000 that Yeargin-Allsopp found in Atlanta in 1996. the authors conclude:
We found that the estimated prevalences of the PDDs studied were probably underestimated. Furthermore, the increasing prevalence and incidence rates during the 1990s may well be explained by changes in the registration procedures and more awareness of the disorders, although a true increase in the incidence cannot be ruled out.
So much for Denmark then. I cannot say that I am looking forward to the rest of this debate if this is the standard of evidence employed by Kirby. But I will resist the temptation to fast forward to Arthur Allen. I will do my blogging duty. Speaking of Arthur Allen, he has written about the debate on his blog and is open to comments. Kirby, despite boasting that whenever he writes on the Huffington Post he goes straight to number one, has yet to share his thoughts on the debate with a wider audience.
STOP PRESS Kirby has made the slides from his presentation available on his website. I wont be downloading them just yet. I don’t want to spoil the ending.
to be continued …
… or not. Kev has blogged the debate here and here. So has Joseph, BC and D0’C. And Diva has set it to music.