OK, this one has been beaten to death. I am amazed that it still think that there is evidence of an “epidemic”. This is especially true of those who rely on the California Department of Developmental Services (CDDS) data. These data are so muddy as to be able to hide a real increase or a real decline.
These data have severe limitations as noted before on this blog. They are not “epidemiological” data. They are not a census of those with autism in California. They are a count of who is getting services and this can and does vary dramatically over time and geography.
That said, let’s take a look at how service rates change with time for a given birth cohort. (click to enlarge). [edit: this is the 1984 cohort] This is much as you would expect. Kids start being listed at age 3. The number increases year after year until a plateau is reached. This happens at about age 7 or 8. There is some slope to the curve: additional kids are being added to the roll even after 8 years old.
This is very straitforward and expected. But, what happens over a longer time to this cohort? Click to enlarge this graph.
Ignoring the obviously leading arrow and label for now, it is abundantly clear that something unexpected has happened. A second large increase in the number of clients is observed. Why would this happen? Well, one of the possible explanations is shown by the arrow. In 1997, the “epidemic” was declared. Autism awareness increased dramatically. One possibility is that the 1984 cohort was still in school where people might notice them and identifiy them. This cohort nearly doubled in numbers from 1997 to 2003.
This brings up so many questions, many of which we just can’t answer with the data we have access to.
It would be interesting to see if there was substitution. Were these kids (heck, teenagers) listed by CDDS under a different label?
How did roughly half the kids in this cohort avoid detection? I think the new phrase is “it’s like missing a forrest fire”. Well, these forrest fires were blazing for 13 years before people started noticing them.
Also, what happened to other cohorts? Well, for one thing, a similar jump in cohort size around 1997 is observed for birth years in the 1980’s and early 1990’s. It isn’t as clear or as consistent birthyear-to-birthyear as you go back in birthyears, but it is observable in some birth cohorts. One example where one can see this is the 1960 birth cohort, which increased about 15% around year 2000.
That last paragrah wasn’t clearly written, I admit. But if you are thinking, “what? The CDDS ‘found’ 15% more 40 year olds?” you read it right.
This graph (click to enlarge) shows the number of CDDS autism clients as listed in 1986 and 2007 by birth year. The 1986 (in black) data are the same as shown before. The drop in the client count in the early 1980’s is an artifact: those kids weren’t identified yet in 1986. The 2007 client count (in red) show something very interesting, at least to me.
There is an increase in autism clients for almost all the birth year groups. 40-year olds, 50 year-0lds, even older people were added to the client list as “autistic”. Again, we don’t know if or how these people were classified before the “epidemic”. They could have been (and likely were, in my opinion) clients listed in another category in 1986.
Let’s take a look at the difference between these two curves. I included data for people with birthdays in the early 1980’s, but these are not reliable. Those people weren’t through the first round of identification by 1986.
The graph shows the difference as a percentage increase. This allows us to see the older cohorts easier. At the same time, it allows people to accuse me of doctoring the data to make it seem like a bigger effect than it really is. That would be an obvious way to try to divert attention from the fact that the “epidemic” caused a roughly 40% increase in CDDS autism clients born in the 1960’s. For those clients born from 1940-1955, the increase was 70+/-28%.
Think about that a second. Autism amongst forty year old people increased by 70% during “the epidemic” years.
How can this be? How could CDDS have missed people with autism for forty or fifty years? Sure, some of these people may have moved into the state. Some of them may have been cared for by family and not been served by CDDS. The trends of these birth cohorts with time do not show the sharp rise in the late 1990’s as observed above for 1980’s cohorts. For me, this is suggestive that the those who could be identified in the school systems, were.
Obviously there are a lot of open questions here. How and why these increases were observed is a big question. Why no one has seen fit to mention this before is another question. The CDDS did not create these data sets for me. Someone else has been paying for that for some time, according to Mr. Kirby…who also hasn’t mentioned this.
People keep saying, “you can’t have a genetic epidemic”. Well, you can’t have an epidemic of a childhood onset “disease” in forty-year-olds either.
____________________________________________
Edit:
First, note that the birth cohort in the first figue above is from 1984. That is not clear. Second, here is a graph with multiple cohorts. Note that all the cohorts have an upswing in the client-numbers in the late 1990’s. Even the 1990 cohort does this. It does not appear to be based on age, but on calendar year.
Left Brain Right Brain 
This shows why I used the 1986 data for this post:
What if you could prevent autism?
The dip in the 1970’s data is not as clear for the more recent dataset as the 1986 data used for that post.
Go ahead and call me “intellectually dishonest” for that if you see fit. I don’t think it is a problem, since I see the increase in the numbers for the 1970’s cohorts to be due to the increased awareness of the 1990’s. I admit that it is rather circular logic.
Also, if you were to fit the data before and after the “bump” in the 50’s and 60’s to an exponential or other growth curve, you would see the “bump” is still there even in the 2007 dataset.
It would be good if someone with access to better (real) data could check on that. I don’t see it happening, though.
Sullivan,
I think you need to provide a bit more data to discredit a recent epidemic.
If your hypothesis is correct that increased awareness around the epidemic declaration year of 1997 would mean that the increase should be visible in all the cohorts at the same time and not staggered based on age. You didn’t really provide us a graph of that data, so it’s impossible to tell if that’s true.
I don’t think your third graph can give us any useful information unless it takes into account the demographic changes by age cohort in the area being studied. Without that data, such a percentage calculation is meaningless and we can all guess whether the increase is due to population changes, or increased awareness. The other thing to check is if they closed a state institution around some of those time periods.
It also would have been helpful to use a different year from 1986, because then we could see data for multiple cohorts who were still in school when the epidemic was declared.
This may be neither here nor there, but I do know there are adult spectrumites receiving services via the _California Department of Rehabilitation_. Additionally, some whom I know of, received extensive services from the Dept of Rehab beginning when they were children. These would be many of the same services which ASD kids receive via the Regional Centers today. I have often wondered if some clients transferred over from the CA Dept of Rehab, to the Reg. Ctrs. within the CDDS.
Schwartz,
I don’t know if this is what you mean, but there is an increase visible in birth year cohorts 1984, 1985, 1986 and 1990 (and probably others. Those are the ones I hae checked).
The 1990 cohort didn’t get a good initial plateau before the increase was observed in the late 1990s–as you would expect. I haven’t checked the others around then. 1980 and earlier don’t show this sudden increase in the late 1990s (I checked every five years back to 1960).
Your comment on the “third” graph. Do you mean the fourth? It doesn’t give any more information than the previous graph, just shows it in a different form.
FWIW the birth year cohort data as surveyed in 1986 would be more likely to tell us about real changes, because it probably doesn’t have an increasing recognition confound.
this is kind of off topic and maybe I am just a hick and don’t understand much. but everyone was thinking that it was the mercury in the vaccines that caused autism. I remember this because my kid is older and falls into that group that got all their shots last decade. so everyone was figuring that autism would drop big time if mercury was taken out. And from what I understand it was taken out a few years ago. but now I am reading that people are saying that the flu vaccines still have mercury. but there should have been a real drop in numbers anyway. Because if you consider the mercury hypothesis for a moment, then they believe that each kid has a different threshold. mercury was in vaccines before the 1990’s but the schedule increased so the mercury increased. so lets say if they are right, more kids became autistic. so now that less, way less, mercury is given in the shot schedule, autism should go down as less kids reach a threshold. Right? I’m confused now by what they were saying when my kid was little, the biomedical community, and what they are saying now. One way or another the numbers would have to decrease if it was the mercury all along. now I read that some people are saying lead also causes autism, but if that is the case then why wasn’t there a huge increase during the lead paint era? it’s not like heavy metals just appeared on the earth at the end of the 20th century! Some people claim that autism started in 1930’s but like I said, it is not like mercury wasn’t around before then. i feel dumb, i’m trying to ask a question and it isn’t coming our right at all. I have gotten some flack from other moms in the autism community over this question and i was also told that my NT kids could become autistic up to a year after their kindergarten boosters by an antivaccine mom.
Erica: What they were saying then was this…
Generation Rescue believes that childhood neurological disorders such as autism, Asperger’s, ADHD/ADD, speech delay, and many other developmental delays are all misdiagnoses for mercury poisoning.
What they are saying now is this…
We believe these neurological disorders (“NDs”) are environmental illnesses caused by an overload of heavy metals, live viruses, and bacteria. Proper treatment of our children, known as “biomedical intervention”, is leading to recovery for thousands.
You gotta give them credit for adjusting their views in light of reality.
Interesting stuff. Schwartz is right, though. If increased awareness caused a spike in reported cases after 1997, it should be visible among most cohorts.
Another interesting feature in the one chart is the huge run up in reported cases starting in the early 70’s. I realize that these are not adjusted for population, but we see the number jump from 100 in the early 70’s to 200 in 1980 to 400 in 1986. It would be interesting to see these in rates per 100,000.
I realize that there could be multiple reasons for the run up, but it is interesting.
If increased awareness caused a spike in reported cases after 1997, it should be visible among most cohorts.
There is an increase in all cohorts. For example, did you see my post on the 18-21 cohort?
I would not expect, say, 50 year-olds with a category of MR in CDDS to have their evaluation adjusted to show autism in order to get autism-specific services. Naturally, increasing recognition will not catch up fully in all age cohorts.
Again, given the limitations of the CDDS data, you can’t make any strong statements. Well, the only strong statement you can make is not to use CDDS data for declaring an epidemic.
That is the point to me. The fact that the count went up for older kids could have a lot of explanations. Mercury causing autism isn’t likely one of them.
Greater awareness is a possibility. In my mind it is a strong possibility. If greater awareness is pushing the numbers for 13 year olds up, surely it would have a big effect on even younger kids.
Also: The tools for diagnosing autism in adults are not as valid or reliable as those for children. So it is very difficult to take an accurate count of how many autistic adults there are today.
Count me as one who agrees with a very minority view among a handful of autism researchers who laugh at the entire ‘epidemic’ or ‘better diagnosis’ discussion. They point out that the introduction of DSM-III-R and DSM IV have not broadened the diagnostic criteria for a PDD, they have expanded it to the point of near meaninglessness. For example Down’s Syndrome people, who would never qualify for a PDD diagnosis before 1986 are now routinely are place on the ‘Spectrum’.
You have people self-diagnosing themselves as having Asperger’s Syndrome by taking an Asperger Quotient Test, which is nothing more than a personality test, setting up web sites and selling their beliefs just as strongly as the promoter’s of alternative therapies.
The polygenic theorists have redefined what the nature of autism is. If you believe the Simon Baron-Cohen’s of the world, autism is not a pervasive development disorder caused by a disruption in early brain development, it is a psychopathological personality disorder, no matter how cleverly he uses politically correct labels.
RAJ, I agree with you. They have pulled and stretched this definition so far and wide, it is practically meaningless. PDD seems to be a dumping ground for many developmental delays and personality disorders.
I have seen the same self-diagnosed Aspies all over the internet. One standard deviation away from the norm doesn’t make you an Aspie, but 3 will probably get you there.
I have said repeatedly that my child would not have been diagnosed with PDD-NOS 10-15 years ago and he’s already lost his diagnosis. Does he have personality traits on the spectrun? Yes, he does. Has he climbed out of his own world? Most of the time, yes. It is all subjective.
The word “epidemic” has been used to get funding, cause fear and democratize the disorcer so that everyone thinks their child might be affected. It’s the AIDS epidemic all over again. Same tactic.
“The word “epidemic” has been used to get funding, cause fear and democratize the disorcer so that everyone thinks their child might be affected. It’s the AIDS epidemic all over again. Same tactic.”
Which is exactly why I try to debunk the notion. I think that the “epidemic” has a real chance to cause harm in the long run.
Sullivan,
“I don’t know if this is what you mean, but there is an increase visible in birth year cohorts 1984, 1985, 1986 and 1990 (and probably others. Those are the ones I hae checked).”
Sorry I wasn’t clear enough. Was the increase all centered on the same year for each cohort, or centered around the same age (staggered per cohort)?
“Your comment on the “third” graph. Do you mean the fourth? It doesn’t give any more information than the previous graph, just shows it in a different form.”
Sorry again, I did mean the fourth graph (I think I had trouble counting that night). A percentage graph like that pretty much has to account for population changes, or it isn’t going to tell us anything useful.
These comments are not to discourage you, because I do find the analysis very interesting.
After all this discussion though, I thought I would look up the term “Epidemic”. Interestingly, the definition is far from objective (from wikipedia):
“In epidemiology, an epidemic (from Greek epi- upon + demos people) is a classification of a disease that appears as new cases in a given human population, during a given period, at a rate that substantially exceeds what is “expected,” based on recent experience (the number of new cases in the population during a specified period of time is called the “incidence rate”). (An epizootic is the same thing but for an animal population.)
Defining an epidemic can be subjective, depending in part on what is “expected”. An epidemic may be restricted to one locale (an outbreak), more general (an “epidemic”) or even global (pandemic). Because it is based on what is “expected” or thought normal, a few cases of a very rare disease like rabies may be classified as an “epidemic,” while many cases of a common disease (like the common cold) would not.”
This is very interesting, because it does not require an actual increase in incidence. It only requires an incidence that is higher than expected.
If the rate is indeed 1 / 160 children, that is a lot higher than people expected EVEN if that has been the rate all along. That would still be classified as an epidemic.
Out of curiousity, since I understand you object to the term epidemic (and I thought the term implied drastic increase), do you object to the stated incidence rate of ~1/160?
Erica,
From what I’ve read, I’ve never seen lead noted as being correlated or even hypothesized as causing Autism. However, Lead is very toxic and has been shown to be extremely damaging in small doses to young children (causing measurable drops in IQ). Higher doses of lead will cause permanent damage in children and more severe symptoms.
Lead has been around a long time, and has been poisoning people for a long time. I would read up on it, because it can easily still exist in many places. If you are concerned, you can easily get a blood test for your children, and it will tell you how much lead they have been exposed to.
There are some doctors that advocate getting regular lead tests for young children because they are particularly at risk (they crawl around and stick their hands in their mouth, so if there is any lead dust around, they will pick it up).
A couple of short things–
1) the data for the different cohorts take off in the late 1990’s at about the same year. Not the same age. The 1990 cohort, for example, sees a rise at about 1997-1999.
2) the word “epidemic” is a code-word in autism, as I see it. It is a word meaning “increase in autism caused by XXX” where XXX is usually mercury.
3) I expect the real rate is more about 1/100. The 1 in 150 now quoted has a huge spread, state to state and by ethnicity. It doesn’t make sense to me that there are real variations of that magnitude. That there can be some variation, yes. But, why would Alabama have such a different number than New Jersey? If there really is a (what 3x?) difference, shouldn’t someone be studying it?
Sullivan,
1) That data supports your article then. Thanks.
2) Yeah, I agree that’s how it’s used, but the CDC’s usage of the term, may be valid given the broader definition of the rate being unexpected.
3) I totally agree that the variations are troubling and study on that is long overdue. At a minimum it could help direct the effort of research.
I added some information in an edit. The main point is the graph showing that the upswing in the client-numbers happens in the late 1990’s for many cohorts. The timing of the upswing is based on the calendar year, not the age of the cohort.
Doesn’t the upswing start a bit after 1994 actually? It looks like caseload was leveling off at around that point for those born in 1984 (who would’ve been 10 in 1994) then a new upswing starts. I would speculate that the publication of the DSM-IV is what correlates with the change.