Friday, October 26, 2012

Redefining autism in the DSM-5

I've got an article out in The Conversation.this week on changes to autism diagnosis in the DSM-5. It's part of a series called Matters of the Mind, discussing similar issues across a range of "mental disorders". For those who aren't familiar, The Conversation is an Australian online newspaper with articles written mainly by academics - but edited by professionals to make them readable! It's a great initiative and there have been some pretty stellar articles of late. So I'm excited to be involved.

In the last year or so there has been a glut of studies comparing DSM-IV and DSM-5. I've listed below all the ones I know of, but there may be even more. In the article, I focus on three of these: the Yale study from earlier this year that I covered here and here; the recent Cornell study that I critiqued earlier this week; and another newish study from Autism Spectrum Australia that, unlike other studies, actually looked at DSM-5 in a clinical context, rather than re-coding old data.

We're still waiting for the results of the official DSM-5 field trials. But, to be honest, the more studies I read, the less convinced I am that we can predict DSM-5's impact in advance. The obvious problem is that there is no objective test of whether or not someone is autistic, so it's difficult to say whether DSM-5 is an improvement on DSM-IV. But much will also depend on the interpretation of the new rules by clinicians; the reaction of governments and insurers to resulting changes in autism numbers and to new diagnoses such as Social Communication Disorder; and ultimately the extent to which an autism diagnosis continues be a stepping stone to services. These are all in themselves difficult to predict.

There is also the broader question of whether it makes sense to base research and clinical practice on the notion of an autism "syndrome". We have the strange situation where, for example, someone can be denied support for social and communication difficulties because they don't have sufficient repetitive behaviours for an autism diagnosis. Or, as one commenter noted, a child with specific language difficulties is unable to access speech therapy that is available to an autistic child without language difficulties. From the research perspective, we're still all just relying on an intuition that the behaviours that define autism actually constitute a coherent syndrome. As scientists we should be always challenging our intuitions but in effect the DSM serves only to formalise and reinforce them.

Further reading:

Emily Singer has a comprehensive review of DSM-5 issues as part of a free supplement in Nature.


Frazier TW, Youngstrom EA, Speer L, Embacher R, Law P, Constantino J, Findling RL, Hardan AY, Eng C (2012). Validation of proposed DSM-5 criteria for autism spectrum disorder. Journal of the American Academy of Child and Adolescent Psychiatry, 51, 28–40 

Gibbs V, Aldridge F, Chandler F, Witzlsperger E, Smith K (2012). Brief Report: An Exploratory Study Comparing Diagnostic Outcomes for Autism Spectrum Disorders Under DSM-IV-TR with the Proposed DSM-5 Revision. Journal of Autism and Developmental Disorders, 42, 1750-1756.

Huerta M, Bishop SL, Duncan A, Hus V, Lord C (2012). Application of DSM-5 Criteria for Autism Spectrum Disorder to Three Samples of Children With DSM-IV Diagnoses of Pervasive Developmental Disorders. American Journal of Psychiatry, 169, 1056-64.

McPartland JC, Reichow B, Volkmar FR: Sensitivity and specificity of proposed DSM-5 diagnostic criteria for autism spectrum disorder. Journal of the American Academy of Child and Adolescent Psychiatry, 51, 368–383.

Matson JL, Kozlowski A, Hattier MA, Horovitz M, Sipes M (2012). DSM-IV vs DSM-5 diagnostic criteria for toddlers with Autism. Developmental Neurorehabilitation, 15, 3, 185-190.

Mattila ML, Kielinen M, Linna SL, Jussila K, Ebeling H, Bloigu R, Joseph RM, Moilanen I (2011). Autism spectrum disorders according to DSM-IV-TR and comparison with DSM-5 draft criteria: an epidemiological study. Journal of the American Academy of Child and Adolescent Psychiatry, 50, 583–592.

Mazefsky CA, McPartland JC, Gastgeb HZ, Minshew NJ. (in press). Comparability of DSM-IV and DSM-5 ASD Research Samples. Journal of Autism and Developmental Disorders 

Worley JA, Matson JL (2012). Comparing symptoms of autism spectrum disorders using the current DSM-IV-TR diagnostic criteria and the proposed DSM-V diagnostic criteria. Research in Autism Spectrum Disorders, 6, 965-970.


  1. The DSM5 Autism Spectrum Disorder is an effort to simplify a complex group of symptoms that frequently occur together in various combinations. It is not derived from research based evidence. There is no legitimate reason to exclude intellectually or cognitively disabled, severely autistic persons from an autism diagnosis as the DSM5 does.

    Catherine Lord has acknowledged in the past that autism research all too often excludes low functioning autism. See "Social Policy Report, Autism Spectrum Disorders Diagnosis, Prevalence, and Services for Children and Families":

    "However, research in ASD has tended to use overwhelmingly White, middle to upper middle class samples, and has often excluded children with multiple disabilities and/or severe to profound intellectual disabilities".

    A good example of such exclusions are the fMRI studies. Parents do not have the option of excluding our low functioning autistic children from haircuts and dental work solely because our children react to the sensory and anxiety challenges those situations present. "Autism Researchers" can not exclude them either and then reinvent autism by re-defining the problem ... our severely autistic children ... out of the official definition of autism. Shame on all who try to do so.

    1. Hi Harold

      Thanks for commenting. As you point out, it's possible to interpret some statements from Cathy Lord and others as suggesting that DSM-5 will exclude autistic people with intellectual disability. I don't think this is the intention (it would be a strange intention given that these individuals are in need of greatest support) and I haven't seen any evidence to suggest that this will actually happen in practice.

      As for your other points about research focusing too much on particular subgroups of the autism population, I agree 100%.

    2. A problem with functional MRI studies in particular is that is you move more than 3 millimeters (the width of two pennies stacked on each other) during the scan, your data is essentially unusable. This is a difficult task even for healthy adults, especially if the task requires your participant to be awake and attending to the things being projected over headphones or on the projector screen.

  2. Hi Jon;

    Researchers from Australia have also weighed in on the side questioning the accuracy of DSM5 diagnostic criteria. Gibbs et al found that about 20% (26/111) of children meeting diagnostic criteria under DSM-IV-TR would not meet diagnostic crtieria under DSM5 proposed criteria. This is a huge problem:

  3. If you believe there is a true underlying condition, autism, which you either have or don't have (albeit with varying clinical manifestations), then you could use a Bayesian latent class model to estimate the sensitivity and specificity of different diagnostic criteria. In principle this is not different to applications of such methods for the comparison of different diagnostics for infectious diseases in the absence of a gold standard (eg ).

    If, on the author hand, you believe autism to be a human construct that doesn't correspond to any single meaningful biological phenomenon but instead reflects historical accident, administrative convenience, and a fatal tendency of humans to categorize everything, even things that don't fall neatly into distinct categories (perhaps a trait selected for over millennia due to the huge evolutionary advantage in being able to accurately distinguish between different species of animals and plants, which of course do fall into distinct categories), then the whole DSM5 exercise seems stupendously non-interesting from a scientific point of view, though of course has huge potential to have a profound impact on many people’s lives. Whether the impact will be good or bad, I don't know, and it's not clear to me how that will even be evaluated.

    My personal view: bin the whole thing and make access to services dependent on need, as assessed by clinical judgement, not arbitrary diagnostic criteria.

  4. I just discovered your blog,and I don't know if you read comments on older posts,but this goes back more to your older post about genetic tests for autism,and just what "autism" itself is?

    I would like to know what you think about cases of autism,that have been proven by a legitimate psychiatric,or developmental psychological diagnosis,but eventually go away,either all or most of it,after treatment for an underlying disease is corrected.This was what happened in my case.I had a diagnosis in between Asperger,and classic autism,with a lot of other co morbidities,like learning disabilities,and serious behavioural problems,like eloping,and self abuse.I also have a lot of serious medical problems.Three years ago,I was found to have a lot of serious problems with B12 and folate.I started,as an adult,on increasingly higher doses of leucovorin.In September of 2012,I had the autoantibody tests,and found I did indeed have cerebral folate deficiency.

    The point is,after three years on leucovorin,everything that got me an autism diagnosis is gone,or almost gone.No one knows what per cent of diagnosed cases of autrism,are made up by these types of reversible diseases.Channelpathies,neurometabolic diseases,what have you.

    This is a point,and a type of disease that is that both more mainstream autism researchers,and metabolic doctors seem to want to ignore.

    A think a lot of families would like an answer as to if these diseases are or are not really autism.The science is so new,that a lot of researchers,doctors,schools,ands other providers of services have not caught up with it.