Wednesday 20 March 2013

1 in 50 children with parent-reported autism in the US

An estimated 1 in 50 US children aged between 6 - 17 years old present with an autism spectrum disorder (ASD). Attention-grabbing isn't it?

Today's post is based on the source of that soundbite, the publication by Stephen Blumberg and colleagues* (open-access) describing results from data mining of the 2007 and 2011-2012 US National Survey of Children's Health (NSCH) relevant to the numbers of cases of ASD.
Testing 1, 2, 3 @ Wikipedia  

The autism-numbers game is something that has been discussed previously on this blog; most recently with the CDC 1 in 88 estimate in mind (see here) and the even more recent data from New Jersey (see here). The direction of the figures seems only one way - up - but the reason(s) for the increase still remain the source of discussion.

The Blumberg report is open-access, but a few pointers might be useful:

  • Based on 91,642 telephone interviews in 2007 and 95,677 interviews between 2011-2012, parental reports of receipt of an ASD diagnosis (autism, Asperger syndrome, PDD) in offspring were noted, alongside other variables such as age of the child, severity of presentation (mild, moderate, severe) and aged when first diagnosed.
  • Actually in the age range 6-17 years old, data were collected from 63,967 interviews (2007) and 65,556 (2011-2012).
  • Results: "based on parental reports, the prevalence of diagnosed ASD in 2011-2012 was estimated to be 2.00% for children aged 6-17". This compared with 1.16% or 1 in 86 for 6-17 year olds in 2007.
  • The increase in prevalence was noted across the age ranges when they were sub-categorised and perhaps not surprisingly, there was a greater increase in prevalence in boys (2007: 1.8% vs. 2011-2012: 3.23%) than girls (2007: 0.49% vs. 2011-2012: 0.70%). Reported severity also shifted between the various data points (and age ranges) indicating that there was a trend towards less severe presentation (milder ASD) post 2008 diagnosis.
  • The authors were able to some degree, rule out "survey-based measurement error" as being a major contributor to the prevalence increase and there is some discussion about the data not necessarily reflecting "factors that exist prior to or occur just after birth". Indeed the authors very firmly suggest that the changes are a consequence of either "recognition of ASD by health professionals or survey-based measurement changes over time". They also conclude that increases in the prevalence of parent-reported ASD especially for children aged 6-13 "was the result of diagnoses of children with previously unrecognized ASD".
  • As per the report on this story in USA Today "15% to 20% of children who were once diagnosed with autism no longer have the condition". Which raises similar questions as to that of the Fein study on 'outgrowing autism' and 'optimal outcomes' which created so many column inches recently (see here and here).

Bearing in mind that this was a prevalence study not an incidence study (see here for the difference) and issues with regards to response rates (2007: 46.7% vs. 2011-2012: 23.0%), and the sole reliance on parental judgement of variables like severity, the data being presented are indeed stark.

That the rates of autism have seen an inordinate shift from what was once considered a rare condition to something which theoretically should [almost] appear in every school classroom at least once is an eye-opener. As mentioned, the debates rumble on about factors such as better awareness of autism, better case ascertainment, diagnostic switching and broadening, etc. as being the source of the increase. A real increase in cases? Hardly a mention in this latest data. The implication that for example our screening methods and skilled professionals have been able to miss or mis-diagnoses a staggering number of children presenting with an ASD is truly mind-boggling and worthy of an inquiry or two in the US and beyond. Indeed better be quick with that investigation with the DSM-V revision deadline fast approaching and the potential impact that might have on the autism numbers game (including adult numbers**).

So 1 in 50 children with an ASD. What happens next? Sure, many children were described as falling into the mild and moderate ability ranges but as I've said before, terms like 'high-functioning' don't necessarily mean 'can function' with regards to daily living skills, quality of life and onwards translating into positive outcomes in adulthood. Certainly society has to play its role in helping people with autism reach their potential (I have a post scheduled on job interviews and autism coming up soon) and changes are indeed on-going (e.g. the implementation of the Autism Act here in the UK). But let's not be too proud of the achievements done in this area, as still many people on the autism spectrum, their families and concerned others have to fight daily for appropriate recognition, provisions and services.

Another important issue also springs to my mind on the basis of the new prevalence figure. Comorbidity, of which autism is by no means immune from, are not mentioned. The realisation that autism is often very much more than the sum of its triad - soon to be dyad - in terms of comorbidity must surely factor into the potential impact of the latest figures. Not least because of the quite startling health inequalities which seem to be present when an ASD is diagnosed and how as was very recently detailed in the CIPOLD report, such inequality can in some cases, have the most profound and far-reaching effects. I'm not trying to scare anyone; just sayin' that we need to be mindful of the whole person not just their autism.


* Blumberg SJ. et al. Changes in prevalence of parent-reported autism spectrum disorder in school-aged U.S. children: 2007 to 2011–2012. National Health Statistics Reports. 2013: 65.

** Wilson CE. et al. Comparison of ICD-10R, DSM-IV-TR and DSM-5 in an adult autism spectrum disorder diagnostic clinic. J Autism Dev Disord. March 2013.

---------- Wilson, C., Gillan, N., Spain, D., Robertson, D., Roberts, G., Murphy, C., Maltezos, S., Zinkstok, J., Johnston, K., Dardani, C., Ohlsen, C., Deeley, P., Craig, M., Mendez, M., Happé, F., & Murphy, D. (2013). Comparison of ICD-10R, DSM-IV-TR and DSM-5 in an Adult Autism Spectrum Disorder Diagnostic Clinic Journal of Autism and Developmental Disorders DOI: 10.1007/s10803-013-1799-6


  1. This is a very interesting and recurring topic. On the web, people always use the ever widening US definition of autism. I have some data, that I plan to dig up, that was very revealing. While there is talk of an autism epidemic in the US, there has been a, rarely commented on, marked decline in mental retardation over the last 25 years. What do you think you get when you make a chart of the prevalence of more than mild autism and add mental retardation? Then you do not see a mental health epidemic. There is an epidemic of diagnosis in the US. My sister is a GP in England, but she does not seem to have very many patients with autism. It must be very hard on parents of severely autistic children when they read about cases of “recovery”; a border-line child has improved and is now “cured”. So now, if they can’t fix their non-verbal and possibly aggressive child, they have somehow failed.

  2. Thanks Peter.

    I agree that diagnostic substitution is one variable potentially contributory to the numbers being described. Having said that, like most things where autism is concerned, there are no doubt other variables similarly acting on the numbers of cases, including it has be said, the possibility of a real increase as at least partially being involved.

    As for differing outcomes, well, again I think it part and parcel of the fact that autism is more likely the autisms, and for some children, with the right support, the outcomes can be very favourable (as per the Deborah Fein data recently).

  3. Maybe it is just me but something doesn't seem quite right with this paper.

    For example, the frequencies table published by the CDC gives a breakdown of the age when a child first was given a diagnosis of autism (K2Q35A_1). I think it includes all children who ever had a diagnosis instead of just those who currently have a diagnosis but the pattern is shows is what I would expect - the majority of diagnosis happen between the ages of 2 to 8 (74.8%)

    Other than a strangely high point at age 18, the graph shows a decreasing number of diagnosis per year. And if you exclude that one point then you get that about 77% of of the diagnosis happened between 2 and 8.

    Which is what I would expect but then the headline of the paper is that a major chunk of this almost 1% increase is from older children getting diagnosed later in life. I don't see how we can get from 3 out of 4 child being diagnosed by 8 to showing a 75% increase that is attributable to children getting diagnosed later in life.

    A simple estimate would be that 1.5% of the total 2.0% number would have been from children 8 and under which would still be a substantial growth the last estimate in 8 year olds of 1.1%.

    I guess it is time to download the raw sas data and spend some quality time analyzing data...

  4. And of course the data that I was looking at for the table had years and months combined (17 could equal 17 years old or it could mean 17 months) so maybe the numbers I cited about age of diagnosis aren't quite right...

    So maybe just ignore me.

  5. Sorry, one last comment. I finished pulling the original data and I stand by my original comment.

    When looking across the entire data set for children who currently have an autism diagnosis in 2012 the breakdown looks like this -

    61% were diagnosed before the age of 6
    25% were diagnosed between 6 and 9
    14% were diagnosed after the age of 9

    For 6-17 year olds the breakdown was as follows -

    55% were diagnosed before the age of 6
    29% were diagnosed between 6 and 9
    16% were diagnosed after the age of 9

    Or, in terms of the 2% prevalence and the entire population -

    1.23% were diagnosed before the age of 6
    0.30% were diagnosed between 6 and 9
    0.27% were diagnosed after the age of 9

    And the 6-17 age group -

    1.10% were diagnosed before the age of 6
    0.58% were diagnosed between 6 and 9
    0.32% were diagnosed after the age of 9

    But in the 2007 data set there is no data about the age diagnosis so entire point is somewhat moot. However, most of the current diagnosis in the 2012 population (81.5%) happened at age 8 or earlier which tends to void the headline that most of the increase happened in older children.

    1. Thanks MJ. Your analysis goes much deeper than mine did for sure and I would encourage you to post about it (if you haven't already contemplated this).

      What I continue not to understand is why there is a dearth of 'incidence' studies over prevalence in terms of how the numbers pan out.

    2. I have been avoiding posting for a while now because I don't have the time to devote to it that I used to have but I decided to take your advice. If you don't mind a small self plug, here it is -

      I think the second chart pretty much tells the whole story. The increase is due to later diagnosis in the oldest children (13 to 17) but the younger school aged children have an equally high prevalence without the later diagnosis.

    3. Thanks MJ, I'm just reading it now.... and very interesting indeed.

  6. Thanks Roger.

    I agree that the gains in early diagnosis are improving the whole diagnostic picture (and potential for implementation of early intervention in whatever shape or form this takes).

    I still have my worries about diagnosing too early insofar as the relative instability of the diagnosis (particularly the sub-diagnoses) although there is a case for intervene first, diagnose later...

  7. "The implication that for example our screening methods and skilled professionals have been able to miss or mis-diagnoses a staggering number of children presenting with an ASD is truly mind-boggling and worthy of an inquiry or two in the US and beyond."

    I don't think it's surprising at all. In the absence of a formal screening program (like we have for breast cancer), diagnosis of any disease will depend on people (parents in this case) turning up and trying to get a diagnosis.

    I think a large part of the recent rise in ASD cases is due to changes in parent behaviour and attitudes. And also to some extent teachers.

    It's very difficult for a professional to diagnose a child with ASD even if they're a textbook case - unless the parents cooperate. Parents can even make it hard to diagnose adults (if they refuse to do the ADI for example.)

    1. Thanks for the comment.

      I would agree partially with your observations in terms of attitudes changing to recognising the wide spectrum of autism.

      In terms of a formal screening program, I would perhaps differ in the conclusion. We do kinda formally screen for autism & dev. issues - at least in the UK - based on things like health visitor developmental checks & the reason for the CHAT, M-CHAT being developed, the guidance provided in every child's yellow book (birth-5 years) and the various assessments as part of reception and infant school entry; even in some cases in preschool facilities. Some of this was included in the NICE guidance on pathways to diagnosis.

      Yes, one could see how for adults born prior to these steps being taken, it would be easy to 'slip through the net'; that alongside the diagnostic substitution that also undoubtedly contributes to the current rise in cases.

      The literature on parents recognising potential issues with autism actually suggests that parents are quite good at spotting developmental red flags:

      It seems however that accessing formal diagnostic services is where the stumbling blocks begin to appear:


Note: only a member of this blog may post a comment.