I should start this new post by explaining why I’ve been mostly absent for the last month. September was an interesting and challenging month. I could say it was a perfect storm, combining the start of the academic year, preparing a new graduate seminar I had never taught before, transitioning into my new role as Assistant Professor at the University of Michigan, and rushing to get my new research laboratory functioning. Fortunately however, things are finally settling down and I can start to catch up with the child psych literature. I hope from now on I will be able to write a new post every Monday morning.

I thought it would be appropriate to start the month with some thoughts about the new Autism prevalence study that was just published in Pediatrics and which hit the news cycle this morning.

In sum, the study found that the prevalence of parent-reported rates of autism was higher than previous estimates. Specifically, they found that 1.1% of all children aged 3 to 17 had autism (1 in 91) as compared to previous estimates of approximately 1 in 150.

Does this suggest an increase in autism rates between 2003 and 2007? Maybe, in that the finding is consistent with the possibility that autism rates are increasing. However, there are a number of issues that must be taken into account, especially the methodological differences between this study and the CDC study that reported prevalence rates for 2002.

For a more detailed explanation of the CDC report from where the 1 in 150 rate comes from read my previous post on the topic: Autism rates in the USA: where did the 1 in 150 number came from?

How was the current study conducted?
The study include an analysis of the 2007 National Survey of Children’s Health (NSCH). The NSCH consists of a random and representative telephone survey of parents of 78,037 children. Each family provided data on a single child, so that if a family had more than one child, only one of the children was selected as the target child for interview purposes. During the interview the parents were asked whether “they had ever been told by a doctor or other health care provider that their child had ‘autism, Asperger’s disorder, pervasive developmental disorder, or other autism spectrum disorder.’ “ If the parent said yes, then they were asked whether the child currently had autism or another ASD.

They found a prevalence of 110 cases of current ASDs per 10,000 children, or 1 in 91 children. As expected rates were 4 times higher in boys than in girls.

There are two key differences between this study and the previous CDC report:

  1. Unlike the CDC report, this study was nation-wide and included all US regions. In contrast, the CDC report was based on autism rates observed in only 15 states. Therefore, the current report appears to be more representative of the US population than the previous study.
  2. The current study was based on parental reports during a phone survey with no corroboration of the diagnoses endorsed by the parents. That is, we assume that parents are correct when asked… does your child current have Autism or ASD? The authors could not verify the veracity of these parental reports through review of medical or educational records. In contrast, the CDC report was based on a review of medical and educational records, and documented diagnoses were doubled checked by trained clinicians to make sure that there was sufficient evidence for each diagnosis. Therefore, the CDC study was much more conservative in estimating autism rates because diagnoses were obtained from medical and educational records rather than from parental reports.

Clearly, neither the CDC nor the current Pediatric study is near perfect, and they provide only a rough estimation of autism cases in the US. Unfortunately, the very significant differences in methodology between these two studies make it impossible to determine with certainty whether the new rate of 1 in 91 reflects a true increase in autism or is simply a byproduct of different estimation procedures not used in previous studies.

The reference: Michael D. Kogan, PhD,a Stephen J. Blumberg,, PhD,b Laura A. Schieve, PhD,c Coleen A. Boyle, PhD,c, James M. Perrin, MD,, Reem M. Ghandour, DrPH,, Gopal K. Singh, PhD,, Bonnie B. Strickland, PhD,, Edwin Trevathan, MD, MPH,, & Peter C. van Dyck, MD, MPH (2009). Prevalence of Parent-Reported Diagnosis of Autism Spectrum Disorder Among Children in the US, 2007 Pediatrics, 124 (4) : 10.1542/peds.2009-1522

Post to Twitter

Tagged with:

15 Responses to Autism rates in the USA: Thoughts on the new prevalence numbers

    • Hi Liz, I saw the UK study but I haven’t read it carefully. I hope to do so this week and I’ll post my comments here.

      ASAN, you are correct that the prevalence seems to be clustering around 1%. I hope this is a stable pattern and the rates don’t continue to increase.


  1. ASAN UK says:

    There seems, despite the many caveats on each study, to be a clustering around 1% of the population mark, with a 4:1 male to female ratio…

    The UK study that Liz quotes, is going to be used for the planning and delivery of services in the UK for probably the next decade.

    If there are significant problems with it, we need to know now. It notably doesn’t include institutions (ie care homes, hospital and prisons).

  2. Betsy says:

    After years of work in the public schools and now working in assessment development, I think you do an excellent job here of going beyond the the data itself to the collection methods. Data can really only be as reliable as how it’s collected. I’ll be referring lots of parents and professionals to this piece.

  3. I am also very interested in your thoughts on the UK study. We need a similar large scale study to be replicated. Although I have tried to maintain an open mind about the prevalence question I can tell that I have believed prevalence to be increasing because I felt shocked by the findings in the UK study.

  4. RAJ says:

    Prior to 1994 Autism prevelance rates were consistent, between 4-6 per 10,000. In 1994 the APA’s working group on autism vastly expanded the definition of how autism is diagnosed. A great deal of concern up to that time was that large numbers of children with developmental problems with associated social-communication problems were falling through the cracks and did not have a DSM category to neatly fit into. It was felt that these children were not being identified and had no opportunity to be referred to special education and early interventions.

    The first DSM (1980) definition used Leo Kanners core defining feature ‘A pervasive lack of responsiveness to other people’. The DSM edition published in 1987 (DSM-III-R)tried to include larger numbers of children by relegating Kanners definition to one of five isolated symptoms, two of which had to present to qualify for an autism diagnosis and Kanner’s definition was no longer required.

    Trials of DSM-III-R found that autism was being overdiagnosed compared to DSM-III and clinical diagnosis. DSM-III-R included 16 isolated symptoms. The working group responsible, attempted to correct the overdiagnosis of autism, felt that there were too many isolated symptoms (16) and with the introduction of DSM-IV they attempted to correct the overdiagnosis problem by reducing the number of isolated symptoms to 12. Thoughtlessly, they completely eliminated Kanner’s defintion and replaced it with the vague, ambigous and subjective ‘Qualitative impairment of social reciprocity’. Impairments in social-communication behaviors are found in mental retardation, childhood communication disorders, development delays, social anxiety, childhhood reactive disorders, pediatric stroke and others. Schizoid Personality Disorder is indistinguishable from Asperger Syndrome.

    Here is the history of the evolution of DSM definitions of Autism:


    ICD-10 and all the ‘Gold Standard’ diagnostic tools quickly followed the lead of DSM-IV and ICD-10 and currently Kanner’s defintion is not present in any of the diagnostic tools currently in use.

    The twelve isolated symptoms contained in DSM-IV is remarkeable in that none of the twelve isolated symtpoms are specific to autism as defined by Kanner. They are all present in varying degrees to neurologically impaired groups from schizophrenia to stroke patients to Alzheimers patients and the developmental disorders as well.

    There is nothing wrong with the current definition of autism with respect to identifying a large group of children with developmental problems who may now qualify for early intervention and special education services.

    The problem is that the autism label is being indiscriminately applied to the large group of children that Kanner first identified in 1943 who do not meet Kanner’s definition. This has made all epidemiological studies virtually worthless since the definition of autism is more muddled than it has ever been.

    Leo Kanner agrees with me. In 1965 he wrote about another period in time (beween 1943 and 1963) that was notable for an unexpected rise in the prevelance of autism in the US and overseas:


    “Almost overnight, the country seemed to be populated by a multitude of autistic children, and somehow this trend became noticeable overseas as well. Mentally defective children who displayed bizarre behavior were promptly labeled autistic”

    “By 1953, van Krevelen rightly became impatient with the confused and confusing use of the term infantile autism as a slogan indiscriminately applied with cavalier abandonment of the criteria outlined rather succinctly and unmistakably from the beginning. He warned against the prevailing “abuse of the diagnosis of autism,” declaring that it “threatens to become a fashion.”

  5. When my son was diagnosed with autism in 1998 a commonly cited figure was 1 in 500. That was soon changed to 1 in 250, 166, 150 and now what 1 in 100 or 1 in 99?

    Diagnostic definition changes (1994) , increased social awareness, alleged greater autism service availability and now … different methods of collecting data. What will the next explanation be? Assuming you are correct perhaps the question should be what has happened to our scientists that they can not devise a study to determine whether the ever increasing rates of autism diagnoses are real or social constructions?

    Are other areas of psychological and scientific inquiry also afflicted by this … lack of ability?

    Are there any psychologists/scientists who do anything except rationalize the increased diagnoses? Oh yes, of course, they are dismissed as quacks.

    • Hi Harold, just a couple of thoughts. Most scientists are not trying to rationalize these changes. Our role is to examine the possible explanations. One explanation is that changes in rates truly represent changes in the incidence and prevalence of autism. Another possibility is that there are differences in how the studies are conducted that explain the different results. This is not something we just apply to the autism rates debate, but it is a basic procedure we apply to explain discrepant results from two different research studies in all topics. In fact, methodological differences is the 1st factor we look at when we see two studies having different results, as these differences provide the most parsimonious explanation for the discrepancies. You will see such comparisons in all discussion sections of all studies about all topics.

      In my post, I noted that these results may reflect true changes in autism rate, but that we have to note the huge methodological differences between these studies. As scientists we can not ignore these differences

      As to your second question, I do agree with you that scientists could be doing a much better job in following a multinational standardized effort to track autism rates. However, properly assessing for whether changes in rates are due to true changes in prevalence is extremely difficult for a number of methodological constrains. This is also not unique to austim, and it is something the entire field of epidemiology struggles with when examining the rates of most medical conditions.

      RAJ: In premise, I agree with most of the issues you raise. Definitions of autism have changed, and this has contributed to the uncertainties that have led to these controversial issues. However, I am not concerned with the fact that the definition of autism today is different that Kanner’s. Evolving definitions and understanding of conditions is not something unique to autism but the norm in medicine. Our understanding and definition of most medical conditions today do not fit the original definition provided by the first person who observed the phenomena. So changes from the original definition do not concern me at all, as long as such changes are based on new understanding of how the condition works. Whether this is the case in autism is a totally different question.

      Thank you for the thoughtful comments. Nestor.

  6. RAJ says:

    “So changes from the original definition do not concern me at all, as long as such changes are based on new understanding of how the condition works”.

    It should concern you. Lumping all disperate neurodevelomental problems under a single umbrella (autism) which lacks definition has led to some positive results, more children carrying an indescriminantly applied label are being given access to early interventions.

    The negative is if you cannot define a problem, the problem cannot be solved. Every month a new media report is published with the usual headline ‘New Autism Gene Discovered!”. Nestor, no gene specific to autism has ever been identified, not one. Name just one and I’ll be glad to explain why you are wrong, just one Nestor.

    The susceptabilty genes are not autism susceptablitiy genes but rather susceptability genes for an early, long lasting disruption of early brain development leading to a wide spectrum of neurodevelopmental disorders and associated social-communication problems.

    As is the case for the uselessness of epidemiolgy studies, genetic studies that have chased the ghost autism genes are useless until subgroups can be identified and reintroducing Kanner’s original definition for at least Autistic Disorder should be included in DSM-V. Will it happen? No, in fact the working group responsible for DSM-V has even suggested including a normal variant category for Autism which will lead to a perfectly normal child who happens to be introverted rather than extroverted can be classified and referred to special education as suffering from an Autism Spectrum Disorder.

    In my opinion, the researchers who have developed diagnostic tools which they claim are ‘Gold Standard’ diagnostic tools are promulgating ‘Fool’s Gold’ and that would include your esteemed Michigan colleague Catherine Lord.

    How would you explain how a diagnostic tool conceived and co-authored by Cathy Lord, the ADI-R, that was used as the Gold Standard tool used to diagnose Romanian orphans who were abondaned at birth and after being adopted by well functioning English foster parents met diagnostic criteria for autism using the Gold Standard ADI-R tool.

    The orphans etiolgy was presumed to be caused by severe emotional deprivation early in life. If Cathy Lords definition is valid, it provides evidence for the validity of Bettleheim’s theory, which is autism can be caused by severe emotional deprivation and which is confirmed by Professor Lords ADI-R.


    “the Autism Diagnostic Interview — Revised (ADI—R; Le Couteur et al, 1989; Lord et al, 1994) was given by a researcher experienced in its use (M.L.R.) when parental interview reports suggested the possibility of autistic features (see Rutter et al, 1999)”.


    • Raj, you are talking specifically about Autism and I’m talking theoretically (thus my last sentence that said “Whether this is the case in autism is a totally different question.” That a definition changes from its original version does not concern me IF such change is responding to new understanding of the disorder (i.e., the original definition was wrong). Changing of the definition per se is not a problem. Changing from the wrong reasons, or under misguided theoretical reasoning would be a problem. Whether this is or is not the case in autism is a different question. N.

  7. Thank you, as always, for your thoughtful and polite response Nestor.

    Do you not think though that with 15 years of ever changing upward assessments of autism diagnosis that the scientific community should do more than simply apply rigid scientific methodology to the latest such survey, important as that assuredly is, and actually devise studies that can provide real answers to determine whether autism is actually increasing, what environmental factors cause or contribute to autism, and what treatments and cures can be developed?

    And please don’t tell me that they are doing so based on overwhelming funding of genetic based autism research. There are credible professionals including those who conducted the UC David MIND study out of California who have pointed out the 20 to 1 imbalance in favor of funding genetic over environmental autism research and have called for a need to equal out the funding.

    With all respect, it is long past time that the scientific community started taking autism seriously and started conducting research aimed at achieving results that can actually help autistic people and their families.

    • Hi Harold, thanks for the comments. From inside academia I can tell you a take my hat off to those doing autism research. It is hard work, with the same scientific challenges of any other field, but they also have to work inside a political and social climate that is too often openly hostile regardless of what research they do. Nestor.

  8. RAJ says:

    You wrote:
    “Changing from the wrong reasons, or under misguided theoretical reasoning would be a problem. Whether this is or is not the case in autism is a different question”

    You can change the definition of a disorder like autism by misguided theoretical reasoning but for the right reason.There was a great debate in the late 1980′s through the mid 1990′s on defining autism. The argument was between ‘lumpers’ and ‘splitters’ that was debated on the pages of the Journal of Auitsm and Developmental Disorder. ‘Lumpers’ complained that there were a large segment of children who were not autistic but were in trouble developmentaly who were denied access to early intervention or special education. ‘Splitters’ complained that by expandin the definitions the new definitions would impair the usefullness of the diagnostic schemes as a research tool.

    The lumpers won the argument with the publication of DSM-IV (1994). The gold standard tools developed by Lord etal (ADOS and ARI) are based on DSM-IV definitions. Can these gold standard instruments result in giving ‘passkeys’ to children who were denied special education’? Judging from the ‘autism epidemic’ the answer is yes and that is the social policy that was considered. Has the gold standard tools (ADOS-G, ARI-R etc) diluted the usefullness of these tools for research purposes, again the answer is yes when the tools can give cutoff scores to unrelated developmental and neuropsychiatric disorders.

    Here is a partial list of disorders which would never have received an ‘Autism’ diagnosis prior to 1987 and the introduction of DSM-III-R (1987).

    Dystrophinopathies (muscular disorders) (ARI-R)


    Down’s Syndrome (ARI-R)


    Early onset childhood schizophrenia (ADOS, ARI)


    Severe early emotional deprivation (ADOS, ARI-R)


    Specific Language Impairment (ADOS, ARI-R)


    In all of these studies (and there are many more not listed) the diagnosis of autism
    was based on gold standard diagnostic tools (ADOS, ARI-R) based on cutoff scores for an autism diagnosis.

    As an instrument for social policy these instruments have served the purpose of providing a passkey to special education and early intervention. It has also produced the phenomena of a mythical autism epidemic and as a research tool, the instruments are useless.

  9. Neuroskeptic says:

    It looks as though the prevalence of “autism spectrum disorders” as they are currently defined is converging on 1% or more (the latest UK estimate was 1 in 64).

    Personally I consider this to be established fact now, I would be much more interested in finding out more about exactly who these children are and how well they function, rather than counting them.

    Because if 1% of children are actually severely disabled by ASD that’s one thing, if most of them are fine just a little bit quirky, that’s quite another.

  10. henrysdad says:

    Autistic kids need help, but this media sponsered increase in autism? Autism skyrocketing? Higher rates? See the YOU TUBE video: autism spectrum seems out of control and “autism epidemic rooted in abuse and misuse of word ‘autistic’ to get a better understanding of what’s going on pretty shocking

Leave a Reply

Your email address will not be published. Required fields are marked *


3 + two =

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Looking for something?

Use the form below to search the site:

Still not finding what you're looking for? Drop a comment on a post or contact us so we can take care of it!

Set your Twitter account name in your settings to use the TwitterBar Section.