Thursday, September 06, 2007

High Prevalence of Autism in Adults

The following is a bullet-point list of pieces of evidence that strongly suggest the prevalence of ASD in adults is high, at least as high as that found in children. [This evidence was first brought together in Autism "Missed" Often, Even Today.]

  • Nylander & Gillberg (2001) screened adult outpatients of a psychiatric hospital and found that 89.5% (17/19) of "definite autistics" did not have a prior autism diagnosis. The most common existing diagnosis in these adult autistics was found to be schizophrenia, which was considerably more common than the diagnosis of autism itself.

  • Stahlberg et al. (2004) found that 30% of referred adult patients with ADHD had co-occurring ASD. The prevalence of ADHD in adults is estimated at 4.2% [source].

  • Over 25 years ago, Shah et al. (1982) found that 38% of the adult patients at a mental handicap hospital had impairment in two-way social interaction, which was "very significantly associated with abnormalities of communication and imaginative activities." This confirmed an earlier study of mentally retarded children (Wing & Gould, 1979). The prevalence of mental retardation is roughly 1%.

  • Baron-Cohen et al. (2001) assessed adults with the AQ test. About 2% of those in the group of randomly selected controls scored in the 32+ range. The researchers also interviewed 11 students who scored in the 32+ range and found that all of them met 3 or more DSM-IV criteria. About 60% of the 11 met threshold criteria. The general distribution of AQ test scores was replicated in Japan by Wakabayashi et al. (2004).

It's possible to propose alternative ways to interpret each piece of evidence, no doubt. But taken together, it's hard to argue with the inescapable no-nonsense conclusion: ASD is common in adults, albeit often undiagnosed.

When I first listed this evidence, one response was that perhaps ASD is common in adults, but that it would have to be a "mild" form of ASD. This is the "children are more severe these days" argument. There's absolutely no evidence that could lead us to believe this is the case, first of all. Second, the world doesn't work like that. In any group of people, there will be a wide range of variation of skills and characteristics. There has to be a distribution of "severity" in ASD adults that goes from very obviously autistic to nearly neurotypical. Sure, the shape of the distribution could have shifted, but is there any compelling reason to suspect this is the case? Furthermore, Shah et al. (1982) provides evidence of high prevalence of ASD in adults who are by no means "mild" anything.

Addendum (11/8/2007)

See Interverbal's post about Matson et al. (2007). The researchers found that 50% of a group of adults with intellectual disabilities met criteria for PDD.

Addendum (04/11/2008)

See Kev's post about a study by Dorothy Bishop. The researchers found that 31% of a group of adults with developmental language disorder met criteria for ASD.

Addendum (10/05/2008)

Baron-Cohen has apparently come out with a new study of university students. "In the Cambridge study, seven of 378 maths students were found to be autistic, compared with only one among the 414 students in the control group." Clearly, he has found that the prevalence of ASD among the students tested is at least 1.01%.

Final Addendum (09/22/2009)

A report on phases one and two of the Adult Psychiatric Morbidity Survey (APMS) 2007 has been published. The first key fact of the report states: "Using the recommended threshold of a score of 10 or more on the Autism Diagnostic Observation Schedule, 1.0 per cent of the adult population had ASD." See also the pertinent post at LB/RB by Anthony Cox.


  1. passionlessDrone9/06/2007 1:44 PM

    Hi Joseph -

    A better title might be "High Prevalence of Autisn in Adults with Other Diagnosed Psychiatric Disorders"

    An argument that is frequently made (though I'm not sure if you have ever made it), is that the unseen autistic adults are out there, but simply never diagnosed as having psyciatric disorder at all. Clearly these studies would not support such a claim.

    Only Baron could have something approaching this, since he had normal controls; but even still, an 'around 2%' value means there was some rounding going on unless 3.5 people scored 32+. Either side of the rounding seems likely to be within the range of stasticial noise. The 11 students in Baron's study who met DSM criteria were not from a random sample, thus it doesn't tell us much about the population as a whole.

    - pD

  2. Yeah, that's a clear weakness in Baron-Cohen et al. (when it comes to this argument). Wakabayashi et al. is a bit better in that regard as it found 3% 32+ scorers and had over a thousand controls. However, most of the controls were university students, which could presumably skew the data in Japan. In Baron-Cohen et al. they said university students did not differ from random controls, but the data seems to indicate otherwise if I'm reading it correctly.

  3. Since being autistic is associated with having a difficult birth, and some researchers believe that being autistic causes that difficult birth, isn't it possible that the prevalence is greater in children as the infant/fetal mortality has been dropping. Researchers have even seen a sprectrum of severity, with more severe autism associated with more severe birth problems. I'll admit that I'm interested in this idea (

    I'm not disputing the high prevalence of autism in adults. I agree with Joseph, but I'm saying that even if the prevalence in adults is much higher than we think, there could have been some rise in prevalence historically, or some rise in the number of more severely affected children. We don't want to mistakenly see this as only two options, epidemic or no change (as I think many people, not necessarily joseph, do).

  4. Donna - It's plausible, but there's just no data on that. We could probably come up with a guesstimate of the effect by figuring out how many children out of 10,000 would've died 30 years ago. Out of those, clearly only a minority would become autistic, although no doubt a considerably higher proportion that in the general population. I'd venture that the effect is not significant.

    The advancing father's age hypothesis is another plausible "real increase" hypothesis. But again, if you look at the numbers reported in the old dads paper, and you consider how much the average father's age has advanced in the last few decades (not a whole lot), the effect should be minimal. (Though, to be fair, maybe average father's age is not the right metric here).

  5. Joseph - I see what you mean. In 1970, the infant mortality rate was 20 per 1000 live births, and in 2000 it was 7 per 1000. I would need better data on how many of those 13 per 1000 could have been autistic in order to argue that this affect could be significant in any difference over the past generation.

    In 1900, the infant mortality rate was 100 per 1000, and a lot of the drop was made by 1950, when it was 29 per 1000. My argument is probably stronger that during that time period, with 71 per 1000 extra children being born in 1950, that it might have caused a significant change in the autism prevalence.

  6. Yes, comparing 1900 to 2007 would probably result in something more significant. (That's really outside the "hidden horde" argument). But there are other factors in the other direction that might have played a part in 1900, such as infections that were not prevented by vaccinations, for example.

    In fact, in 1920 California, about 60 in 10,000 persons lived in institutional care facilities. I don't believe it's that high today. Granted, intitutionalization rates are probably affected by cultural factors as well.

  7. This comment has been removed by the author.

  8. Joseph wrote:

    "But taken together, it's hard to argue with the inescapable no-nonsense conclusion: ASD is common in adults, albeit often undiagnosed.

    When I first listed this evidence, one response was that perhaps ASD is common in adults, but that it would have to be a "mild" form of ASD. This is the "children are more severe these days" argument."

    An alternate hypothesis is that - paraphrasing a comment made elsewhere by Prometheus - the developmental delay often associated with ASD is just that, delay (and/or an alternate developmental path) rather than stasis. As such, some of what might be regarded as the more noticeable and/or disabling aspects of ASD in children may be ameliorated over time via ongoing development (e.g. the non-verbal 4 y/o becomes a verbal adult) and/or learning alternative ways of coping and accomplishing tasks. The result would be a reduction of some of the more 'visible' aspects of ASD, increasing the percentage of ASD adults who could be considered 'mild'.

  9. Hi Ian. That's probably the case. Just to take one example, not all non-verbal children will be non-verbal adults. Many adults currently getting diagnosed with Asperger's probably would've had a more appropriate diagnosis of autistic disorder as children.

  10. I'm sorry, I know this is an older post, but the more I learn about autism, the more I believe I might be on the spectrum.

    While I don't want to condone the "kids today have WORSE autism" idea, it might be valid on the level that, to get school help, it at least seems like more kids are getting labeled with an ASD. This could lead to a skewing of what it means to have 'severe' autism (or, more likely, 'moderate'), with people like myself--never diagnosed, never significantly delayed, but extremely socially awkward and with ritualized thinking patterns--getting diagnosed as PDD-NOS or Asperger's.

    I was never diagnosed; never even considered for it, as best as I know. I have a few stimming behaviours which I learned at a young age were not acceptable and kept private. I was at an utter loss in social situations, but it was chalked up to being just an odd kid with crap social skills. I fixate(d) on a subject of choice (ironically, the current subject is autism) and collected information. I'm uncomfortable with unasked for touch. I don't think I qualify for a diagnosis on the spectrum, save possibly a BAP sort of "Well, you have traits, but not really enough of them..." And I am certainly not alone. If I qualify, my girlfriend does, too--she's similarly and equally awkward. In fact, of the few friends I have, probably half are very similar to me.

    I think that if we were all starting kindergarten in the US today, in a state with very good special education resources for autism, we would all be diagnosed as on the spectrum. And I think that the same is probably true for a lot of young adults in my generation (a 20-30 year old cohort). We were just the weird kids and that was the end of it. There was no awareness in our parents that what made us weird might be something like autism, especially given inaccurate ideas of what autism is/was--in fact, when I brought up the idea of my testing strongly autistic on the Baron-Cohen AQ test and similar sorts of things, my mother balked at the idea ("You're too verbal, too good with words, and not good enough at math!"). Today's parents know about autism. I think there is merit to the idea that we are 'creating' autism through awareness: as more people learn about autism and the range of symptoms it can present, more kids who would have been classed as social misfits or weird a generation ago are instead given a label and IEP. The number of weird, socially awkward kids who don't have classical autism, at the very high-functioning end of the spectrum, is probably the same as ever. They/we were just overlooked before.

    I apologize for my novella!

  11. This comment has been removed by the author.