Monday, October 15, 2007

Gallup & Yazbak: "Prevalence is 1 in 67"

Raymond W. Gallup & F. Edward Yazbak have posted an analysis titled "When 1 in 150 is really 1 in 67." As you can imagine, they did not survey children in order to arrive at a new prevalence number of 1 in 67. They estimated it as follows.

  1. The CDC has reported that the prevalence of autism among children born in 1994 is 1 in 150.
  2. The administrative prevalence of autism among 6 year olds, as reported in IDEA, has increased 124% (factor of 2.24) from 2000 to 2006.
  3. Therefore, they argue, the prevalence reported by the CDC must actually be 1 in 67 for children born in 2000, if adjusted accordingly.

Anyone see why this methodology does not hold?

For this analysis to hold, IDEA ascertainment would have to be at least roughly equivalent to CDC ascertainment. It's not a matter of "not believing the statistics collected by the U.S. department of education" as the authors contend. I believe them. I just don't think they are counting all autistic children, by far. But I do think they are getting better at ascertaining autism.

In 2000, the prevalence of IDEA autism among 6 year olds was 0.23% (source, table B2C). This is 2.89 times less than the CDC prevalence reported for children born in 1994. So for every 3 autistic children, IDEA missed 2. Of course, it's unlikely they missed most of them. They were probably served under other categories, as demonstrated by Shattuck (2006).

Could it also be that the IDEA category can be changed after the age of 6? (Note that in IDEA a child is in one category or another, not multiple ones simultaneously.) The IDEA prevalence of autism among 11 year olds in 2005 was 0.47% (same table), more than double that of 6 year olds in 2000. Obviously, half of the autistic 11 year olds, as recognized in 2005, must have been served in other categories in 2000.

So Gallup & Yazbak's analysis is flawed in regards to using the IDEA prevalence of 6 year olds in 2000 to extrapolate the results of a survey done recently on children who are now older. That's just a demonstrable flaw of the analysis. I don't believe the analysis methodology is valid either way.

21 comments:

  1. The CDC and IDEA numbers are equally rife with errors between years and categories. Most likely 100 monkeys throwing darts at prevalence rates would be more accurate in picking the right rate.

    ReplyDelete
  2. The CDC survey is probably fairly accurate at determining the prevalence of diagnosed autism. It's not a whole population screening, of course, which would give a considerably higher prevalence.

    IDEA doesn't seem accurate at all, but autism is being recognized more and more there in favor of other categories.

    ReplyDelete
  3. I'm just glad that there are people out there able to translate the numbers for me!
    Mathematically challenged person.

    ReplyDelete
  4. School district criteria for diagnosing autis can vary widely----what about regional differences? (Here in New Jersey, awareness of autism is very, very high, so that "E/I" is tantamount to "autism, maybe"). From 2000 when Charlie was 3 till now, how many people know about autism and what they know about it differ greatly. Thanks for the analysis.

    ReplyDelete
  5. This is reading to me as a mathematical calculation based on a false premise. It assumes that ALL children of DXed adults will have Autism. This proves how little we understand scientifically about Autism. The gene can clearly skip generations (in my opinion) and such a thing is not unusual in established genetic conditions.

    Remembering also that it's likely that a number of DX's are wrong - given that the figures cover the whole Autistic Spectrum, and in some places includes ADD and ADHD individually. Those two have a multitude of DX's forced by parents who want an excuse for their child's poor behaviour which in reality is the parents fault.

    ReplyDelete
  6. Kristina; one of my favorite quotes from David Kirby...

    "But the national autism rate within IDEA is only 25 per 10,000 among 3-5 year olds, far below the estimated rate of 60 per 10,000. Ohio has only 9 cases per 10,000, Puerto Rico just 7, and Oklahoma just 4 (about 1/20th the rate of Maine).

    Clearly, we need to let the diagnosing and reporting catch up before drawing any solid conclusions about autism...
    "

    ReplyDelete
  7. Timelord: I'm not exactly sure what your argument is, but if you're suggesting that if the "autism gene" (there's likely no such thing) skips a generation, then you would see a generational fluctuation in prevalence, no, things don't work like that.

    ReplyDelete
  8. Hi Joseph I don't know if this is relevant or not but the 1 in 150 number the CDC produced was just an average with a great deal of disparity between different regions of the country. For example the prevalence rate of autism in New Jersey was more than three fold than in Alabama. The authors imply that mercury in vaccines or just vaccines themselves cause autism and it is not due to differences in diagnosis. If there are much greater differences in vaccine rates in New Jersey than in alabama the authors don't document it. Various states (including Alabama) only had access to health records and not school records. Various states (including New Jersey) had access to both health and school records. There were higher rates of ascertainment in states with access to school records and health records, so i am not sure what if anything the 1 in 150 rate means since it was just an average of different regions and not adjusting to the differences in populations of New Jersey and California for example. So I don't think the 1 in 150 can even be used as a true national prevalence number. If the authors are implying there is a true rise based on environmental numbers and the states don't have huge differences in vaccine rates, they have to show some other environmental difference that just happened in New Jersey that did not happen in Alabama in the late 1980s to early 1990s. The differences between regions and differences to access of different type of records would suggest against an environmental argument.

    Their argument that the schools would not be offering these services to persons with autism unnecessarily is also flawed as it ignores the huge amounts of litigation that take place either with plaintiff prevailing or the costs of litigation exceeding the costs of providing services forcing the schools to settle.

    ReplyDelete
  9. Jon: I agree. If vaccination explained regional differences, it probably wouldn't be hard to come up with a quick analysis to demonstrate this. Differences are probably due to economics, population density, availability of specialists, things of that nature.

    Also, regional differences in IDEA are huge compared to CDC differences.

    ReplyDelete
  10. How did I miss this?

    This math... just wow...

    The fact that you can't mix the different systems should be painfully obvious even to those with only a basic understanding of math

    ReplyDelete
  11. Timelord: I'm not exactly sure what your argument is, but if you're suggesting that if the "autism gene" (there's likely no such thing) skips a generation, then you would see a generational fluctuation in prevalence, no, things don't work like that.

    OK, maybe I over generalised making it look like I was talking about a gene when it's more complicated than that. But I'm interested in your rejection of my theory - given that there appears to be a genetic "carry on", but at the same time there isn't a corresponding rise in number. If you get what I mean there.

    ReplyDelete
  12. Timelord: What I mean is that there are no specific times at which generations start. It doesn't matter if genes skip a generation or not. Real prevalence due solely to genetics should be stable, unless new mutations are introduced (advanced paternal age) or unless there's selection in favor or against certain mutations.

    ReplyDelete
  13. dont you think though that this is all a red herring?
    Whats important is that children need attention to their developmental delays.
    It all comes down to what the definition of autism is.
    If you are talking about the gentetic dispositions (and maybe wiring of the brain) that is passed down from parents to siblings, then i doubt that there is an increase in autistic children.
    (I doubt...)
    But if you are talking about the medical issues that cause autistic symptoms then we should be looking into whether or not there is an increase in these cases.
    But that would pretty much be impossible...here is where anecdotal evidence doesnt help.
    Teachers claim that there is an increase in children with certain symptoms that are consistent with autism, but they might just be more aware of what those symptoms are.
    It's impossible to get any concrete evidence that there are more cases of children with autistic like symptoms than there were 20 years ago.
    But its important.
    If there is a rise in the cases of children with a developmental delay we need to know and find out whats causing it.
    As far as 1 in 150...I never believed that.
    They changed the criteria for what is autism...so of course the numbers will jump when you make it more inclusive.
    Phil

    ReplyDelete
  14. One way is to look at the whole population of children in special education. Or you can look at an aggregate of autism, MR, speech delay, learning disability and emotional problems. There's no increase in children aged 8 or 9 in special ed at least between 1993 and 2005. There's a bit of an increase in children 6 or 7, but this probably reflects earlier diagnosis.

    ReplyDelete
  15. This comment has been removed by a blog administrator.

    ReplyDelete
  16. The previous comment was spam. As promised, anything that violates Blogger's content policy is deleted.

    ReplyDelete
  17. What I mean is that there are no specific times at which generations start. It doesn't matter if genes skip a generation or not. Real prevalence due solely to genetics should be stable, unless new mutations are introduced (advanced paternal age) or unless there's selection in favor or against certain mutations.

    Oh OK - I think get you now. What you're saying is the population at large increases the percentage won't. Right? If so, then I missed that one!

    On what the other Phil said, he's right - the numbers went up when the DSM-IV was introduced in 1994 because it widened the interpretation of Kanner's Autism to include Hans Asperger's work, as well as help in the correct DX of schizophrenia. Somehow I think ADD and ADHD got mixed up in there as well when it really shouldn't have.

    I don't think that's a bad thing to be honest - after all I'd still be in dark about myself if that hadn't happened. Also - contrary to other views - it IS possible to DX HFA and Aspergers in adulthood which would also have an effect on the numbers.

    It's just a shame that we can't achieve posthumus (sp?) DX's. That'll put the wind up the crazies demanding births before 1931! Oh Joe while I think of that - do you have that CDDS data that a blogger on LB/RB referred to a little while back?

    ReplyDelete
  18. Sullivan's data? It's a shame that's lost for now. I'm sure he has the graphs still, and maybe Kev will allow access to archives somehow.

    ReplyDelete
  19. As I recall the post on LB/RB, Joseph, the writer said that both Kev and you had the data. Not just the graphs - that was the way I read it - but a whole lot more. Like the total number of different people born before 1931 who had been listed on the CDDS as on the Spectrum dating back to 1992. IIRC the post mentioned three being on the list now. What interested me above all else is how many in total from 1992 to now were on the list that were born before 1931. If we have a verifiable stat for that AND it's a significant number we can blow the mercury theory right out of the water once and for all!

    We hope! (heh)

    ReplyDelete
  20. That's right. Sullivan, Kev and I have the files from CDDS that David Kirby brought to our attention.

    There are some autistics in CDDS born before 1931. As you can imagine, not many are recorded, but there's no spike in birth year cohort trends around 1931.

    ReplyDelete
  21. Southeast and main Asian pandora jewelry countries have twisted rubies for centuries, cheap pandora bracelets but research as to where, and how to find more deposits is Pandora charms spare, and production has figured out how and mining companies,” Pandora beads Giuliani says, to look at exactly the right time and place.” pandora set Farther investigation of claret formation, based on tectonic scenery, cheap pandora geochemistry, fluid inclusions and isotopic ratios, allowed discount pandora Giuliani’s lineup to remodel a new prototype for the French Institute pandora 2010 of Research for Development (IRD) and the National Scientific pandora sale Center of Research, two government-sponsored knowledge Pandora Bangles and technology research institutes that aim to aid in the sustainable cheap pandora bracelets development of developing countries. Before the collision pandora bracelets prices of the Eurasian and Indian plates, lagoons or deltas sat in the regions where marble is giant, pandora bracelets and charms he says, “and there is the brains to expect that the new pandora bracelets sale thoughts should help development of the artless capital.” discount pandora bracelets Virginie Garnier, Gaston Giuliani and Daniel Pandora necklace Ohnenstetter urban the shape to do just that. They work for the garnet cheap pandora charms genesis. While studying the bedrock in Vietnam in 1998, the discount pandora charms French players found rubies, which detained traces of aluminum, chromium pandora charms sale and vanadium from universities, international corporations, governments pandora charms 2010 and why the rubies got there, and has created a paradigm Pandora beads to help these evaporites, Garnier says, when the Eurasian cheap pandora beads and Indian plates collided, raising the Himalaya Mountains.

    ReplyDelete