Monday, July 19, 2010

Prevalence of Autism in Adults with Severe Intellectual Disability

One of the many unanswered questions when it comes to autism is just how many adults have autism.  While this may seen to be an academic question at first, the answer to the question is quite important.  The reason is that if we knew how many adults have autism it would give us a clue to whether autism is more common now than it was in the past.  If there are just as many adults out there are children, that would mean that autism is not becoming more common and it is likely that we are just better at recognizing it.  If there are not as many then that would strongly suggest that something else is going on.

Unfortunately, there have been very few (if any) high quality studies that have looked for autism in adults.  For example, the most recent study that looked for autism in adults in the UK came up with a figure of 1% but the study had major problems with how it was done.  As a result, we can only guess at how many adults have autism.

With that in mind, consider a recent study published that was recent published in the Journal of Intellectual Disability Research.  I wish I could say that it directly looked at the autism rates in adults but it didn't.  What it did do is look at how common autism is in adults with intellectual disability.

Intellectual disability (ID) is roughly defined as having an IQ under 70 along with potential problems with motor, language, and self help skills.  ID is thought to be fairly common in people with autism - estimates range from 20% all the way up to 80% when you include borderline ID.

In this study, the researchers looked at all of the adults (ages 18 to 67) with severe intellectual disability (IQ less than 50) who were registered for government services in in Reykjavik, Iceland in 2000.  The city had a population of 70,007 and there were a total of 256 adults were registered for services for severe ID.

The researchers sent out letters to all 256 adults and attempted to recruit them for this study.  However, only 119 of the 256 people elected to participant.  Those who elected to participant were significantly younger and more likely to live in group homes than those who didn't.  As you will see, this fact become important later on.

These 119 people were then screened (by psychology students working on their thesis) using a slightly modified Bryson scale followed by a the Childhood Autism Rating Scale (CARS) with modified cut-off point (26 or above instead of the usual 30 or above).  Afterwards, an experienced clinician administered an ADI-R to confirm that the person did in fact have a form of autism.

The overall results were not that surprising.  The study identified a total of 25 cases of autism, 14 of which had did not previous diagnosis of autism.  Seventeen of these cases were autism while the other 8 where atypical autism (think PDD-NOS).  I could not determine from reading the study what the exact breakout of autism vs atypical autism was in the initial 11 cases compared to the new 14 cases.  It appears that the 11 people who already had a diagnosis had a diagnosis of autism which means 6 of the new cases were autism while 8 where atypical autism, but I could easily be wrong.

The breakout of autism vs atypical autism in the groups may sound like a minor point, but it actually can help us determine why the cases were "missed" in the past.  One of the arguments against the rate of autism actually increasing is that the diagnosistic criteria have undergone major revisions over the years.  The earlier criteria were much more limited in what they would call autism while the most recent crieria have expanded their scope, especially in the area of atypical autism or PDD-NOS.  Therefore, if the newly diagnosed group were mostly in the atypical group, it could be argued that they were "missed" earlier because they would not have met the criteria for any form of autism.

But from what I can gather, it is a mixed bag with only about half of the newly diagnosed people falling under an atypical label which means that the change in the diagnostic criteria only played a partial role in the new cases.

Regardless, one of the other interesting points here is that most of the people diagnosed were relatively young.  The ages ranged from 19 to 63 with a mean of 32.56 and a standard deviation of 10.92.  The study does not give more of a break out of how these cases are distributed by age, but from looking at the statitics, you can tell that there are far more younger cases than older ones.  I don't know whether this age bias is a function of the subset of the population that responded (which we know were younger), or whether it is a function of autism becoming more common each year (younger people have a higher autism rate), or both.  I suspect it is both.

Going back to the population of adults with severe ID, the rate of autism in this group was about 21%.  This number is not surprising and, if anything, is on the lower side.  Other similar studies have suggested that the rate is higher, possibly around 30%.

But in general, the results of this study agree with earlier results and doesn't give us much new information.


Saemundsen E, Juliusson H, Hjaltested S, Gunnarsdottir T, Halldorsdottir T, Hreidarsson S, Magnusson P. Prevalence of autism in an urban population of adults with severe intellectual disabilities - a preliminary study. J Intellect Disabil Res. 2010 Aug;54(8):727-35.  PubMed PMID: 20633201.
DOI: 10.1111/j.1365-2788.2010.01300.x


  1. Remember the old saying about looking for a needle in a haystack. I think this rings true for trying to find the true prevalence of adults with autism. The CDC studies which show prevalences of close to 1% only deal with children because children present to services for special education which adults over 21 are not eligible for. Therefore they are easier to find, in addition to the fact that I believe under federal law people can't refuse to participate in CDC surveys.

    Adult populations, since they don't present for services are much more difficult to find. From what I know about that British study, it had some methodological problems and the numbers were just projections based on a base population of 19 people and it was just an estimate based on how various populations did on a nonstandardized test that the number was actually 1% If there were actually some way to survey the adult population and find the true prevalence that would be great, but I don't think it is really feasible for the above stated reasons.

    It would be good if persons over 70 could be found so we could encourage them to donate their brains to science for postmortem autopsies. Then we might eventually find out the brain etiologies of a lot of ASDs.

  2. Hi MJ -

    Thanks for taking the time to write this up.

    Am I missing something, or isn't this a rather spectacular failure to find anything like a 'horde' of people with autism? It looks like around 10% of the population with IQ < 50 fit the criteria of hidden autistics. I just don't see how a finding of 14 people with autism does very much to a city that we'd expect to see 700 autistics in. (?)

    - pD

  3. I'll admit I'm a bit confused, as to whether this report is saying that 21% of those who have IDs are also autistic, OR that 21% of those who are autistic also have ID.

    It sounds like they're saying the former is the case, I'm not sure.

    I would rather believe that 20 to 30% of autistics have ID, that sounds about right to me, but I have no way of really knowing.

    I agree, it's important to know how many autistic adults are out there, not only to know whether we're increasing at a higher rate than the general population, but also to provide needed services.

    I'll agree with Jonathan, it can be hard to find them, as they usually don't get DXd until there's a problem.

    Hey Jon, let's both agree to donate our brains for research!

  4. You are right, Jonathan, it would be harder to count the number of adults with autism than it is to count the number of children - and even with children there are still a the wide variety of estimates. But still, it should be feasable to do some sort of survey or study of adults to get a rough idea, all it would take is the funding to conduct the study.

    I think that part of problem is that the medical community doesn't really want to know the answer. If the number comes back low or shows (as I think it would) a downwards slope with age, then they are going to have to come up with an explanation as to what that means.

    So instead, we get speculation about the "hidden horde" and occasionally a study like this is done that seems to suggest that there are more adults with autism than we thought.

  5. PD,

    I think the point here would be that the percentage change in the population and not the raw numbers. You can look at the numbers and say that there was a 120% change in the population or over half of the population was not identified, and that was in people who were closely examined. So the logic would be that is half the population in this population were missed, just imagine how many are missed in the higher functioning groups.

    The problem is that the numbers don't add up to what they should. Even with these new cases and the assumption that they would be reflected in the general population, you can't easily get to what the expected prevalence would be today. I am working on a follow up post to this that attempts to get into what the numbers would look like.

  6. Clay,

    The figure from the study is that is 21% of people with severe ID also have autism.

    The number of people with autism who also have ID (severe or otherwise) is somewhat of a mystery. The CDC puts the figure at an average of 41% (in children) when you consider all forms of ID (IQ < 70) but that figure is highly variable between the states. When you consider borderline ID (IQ < 80), that figure jumps tremendously, up to almost 80% in some states. Of course these numbers could be inflated because of problems testing intelligence in children with autism but I would suspect that we are still looking at about 50% when you include children with borderline ID.

    As to the DX part, I think that is part of the problem with the story. Children are diagnosed today because they have problems. So either the kids in the past were called something else, their problems with ignored, or there weren't as many kids with autism. While I would assume that the first two reasons did happen, I don't think that they can come close to the increases that we have seen.

    I think you are dealing with a different animal when you have have someone who makes it into their adult years before being diagnosed. A simple example here would be that a huge percentage of children with autism - 40% - do not talk at all. That would be very hard to miss.

    A child who does not talk would not get into their adults years without a diagnosis of some sort and would find it very difficult to function normally until their autism is "discovered" many years in the future.

  7. You said:
    "I think you are dealing with a different animal when you have have someone who makes it into their adult years before being diagnosed. A simple example here would be that a huge percentage of children with autism - 40% - do not talk at all."

    I know it's just a phrase that is used, so I won't take undue offense at the "different animal part of your reply. I know what you meant, and as a person with Aspergers who wasn't DXd until the age of 53, I qualify as such.

    I would inquire where you got that figure, and point out that the definition of "children with autism" you must be using must not include those DXd with Asperger's.

    I think that 40% seems high even for those you would classify as "low-functioning", so that's why I ask where you got it from.

  8. Clay,

    The 40% figure comes from the CDD's data page here -

    and sources the figure from this study -

    Johnson, C.P. Early Clinical Characteristics of Children with Autism. In: Gupta, V.B. ed: Autistic Spectrum Disorders in Children. New York: Marcel Dekker, Inc., 2004:85-123.

    I have not read the study but the CDC has this to say -

    "About 40% of children with an ASD do not talk at all. Another 25%–30% of children with autism have some words at 12 to 18 months of age and then lose them. Others may speak, but not until later in childhood"

    As the quote uses "children with an ASD", I would assume it would include children with Aspergers in the total number of children. Although you are correct that children with Aspergers should be able to talk.

  9. Thanks for providing the link. I don't really know how to access the study (PubMed?). It sounds like they are including Asperger's, but if so, the numbers can't be correct.

  10. I believe this is the link to source, it is a book -

    or on Amazon -

    As for correct or not, well, I don't think that the CDC would put up a number of their main autism data page without having the science to back it up.

  11. And thanks for those links, but at $229.95, I don't think I will be buying it. Yikes!

    I'll try to find out about that 40% though.