The Epidemic That Still Isn’t: Autism Rates and Case Definitions

A paper that came out yesterday in the American Journal of Psychiatry has generated a lot of press coverage of the “autism epidemic,” as it purportedly shows that one in every 38 South Korean children is autistic. That’s more than double the incidence previous epidemiological studies have found. I have no doubt that this result, completely devoid of its larger context, will now be picked up by all manner of woo-woo peddlers as indisputable proof of whatever nonsense they’re selling. What the paper really shows, though, is how slightly different interpretations of case definitions can produce radically different results.

First, let’s dispense with the most obvious flaw in much of the news coverage of autism: there is no “epidemic.” As numerous epidemiologists have found every time they’ve looked at this issue, the apparent rise in the rate of autism spectrum disorder (ASD) tracks perfectly with changes in the diagnostic and special education criteria for the disease. It’s not rising, just being identified more often.

One particularly telling point is that the supposed increase in ASD doesn’t age well. Autism appears in childhood, but many folks seem to forget that children grow up, and the disease doesn’t go away when they do. That means that if ASD rates are genuinely increasing, there should be more autistic kids than autistic adults. But as a paper last week in Archives of General Psychiatry showed, there aren’t.

A Reuters story on that paper explained the key finding nicely:

Researchers found nearly one percent of Britons older than 16 years have autism, a rate that is similar to that seen in children. Younger people were no more likely to be affected than older ones, however, which would have been expected if the condition were truly on the increase.

“It was surprising to all of us,” said Dr. Traolach Brugha, a psychiatrist at the University of Leicester, who worked on the study. “If this study is correct, it does put a big question mark over the autism epidemic.”

Not so much a question mark as a stake through the heart. Brugha’s team found that the adults with ASD were less likely to have been diagnosed previously than kids with it, pretty much proving that the rising rates are due to increased diagnosis and awareness. Nor is Brugha’s report the first in this genre. For example, Orac at Respectful Insolence has an excellent post about a 2006 study that essentially proved the same point using a different method.

So case rates aren’t rising. But are they really only 1%? That’s what Kim et al. wanted to determine in the new South Korean study. I encourage everyone to read the whole paper, which Am J Psych has made available for free. It’s an impressive piece of work.

First, the researchers identified a South Korean community that was demographically diverse, and set out to sample all 7- to 12-year-old kids in that region. There were 55,266 of them. Before I start my criticism of this study, I have to congratulate the authors on the dedication, organization, and plain hard work that they put in to tackle such a huge job. As I said, it’s an impressive paper.

Kim’s team broke their sample into two groups: those who received special education or other psychiatric help (the “high-probability group”), and those who were in the general school population. That ensured that their sample wouldn’t be skewed by accidentally pulling too many kids from special ed. Between the two groups, they found a whopping 2.64% prevalence for ASD, or about 1 in 38 kids. That’s the conclusion that’s been plastered all over the headlines, and the one that I predict will launch a thousand quacks.

As often happens, though, the Abstract giveth while the Results taketh away. Let’s start with the autism rates in the high-probability group, which were, unsurprisingly, very high:

In the high-probability group (those in the Disability Register, those in special education schools, and those in regular schools who had psychiatric or psychological service use) 97 of 114 children were confirmed to have autistic disorder (N=74) or other ASDs (N=23). The high- probability group contributes 0.18% for any ASD to the total population prevalence (autistic disorder=0.13% and other ASDs=0.05%; the ratio of autistic disorder to other ASDs was 2.6:1).

I’ll parse that. In ASD, the middle letter is critical: autism occurs on a spectrum. Contrary to popular belief, not everyone with ASD is Rain Man. In fact, the cases depicted in movies are generally the extreme end of the spectrum, people with the biggest problems. Milder cases extend from there all the way through (according to many researchers) Asperger Syndrome. There are no bright lines dividing these folks. Indeed, there’s no reason to believe that there’s even a clear boundary between “sick” and “well” here. Extremely autistic individuals clearly need help, but what about high-functioning folks with Asperger’s? At what point does one divide someone with very mild ASD from someone who’s just socially awkward? Like most psychiatric case definitions, it’s murky at the edge.

What Kim et al. found in the special ed group was a lot of ASD, with the majority of diagnoses in the “autistic” category rather than in milder categories. These are the sickest kids. Presumably that’s why they’ve been singled out for special help. So far, so good.

The general population data get more problematic:

For 104 children with ASDs in the general-population sample, among the 172 assessed, the crude prevalence for any ASD was similar to that in the high-probability group (0.19%). However, the ratio of autistic disorder to other ASDs was reversed, with prevalences of 0.05% and 0.14%, respectively (ratio, 1:2.6) (Table 2). Other differences between the high-probability and general-population groups included the ratio of boys to girls, which was 5.1:1 in the high-probability group and 2.5:1 in the general-population sample (p=0.037) (Table 2). Mean performance IQ for individuals with any ASD was 75 (SD=28) in the high-probability group and 98 (SD=19) in the general-population sample (p<0.001)

Now we’re seeing a majority who are on the mild end of the spectrum. How mild? In some cases, quite. The surveys the researchers used to assess ASD, while standard in the field, suffer from the usual drawbacks of any psychiatric survey tool. Does your child have only a few friends? Prefer playing alone? Is he or she easily overwhelmed by extraneous stimuli? There’s lots of wiggle room on questions like these, and the closer you look the more likely you are to start pathologizing mere eccentricity.

Finally, there’s an extrapolation problem. As the numbers in the paragraphs above indicate, the researchers didn’t manage to get detailed diagnostic information on all 55,266 kids in the district. A lot of parents responded to the initial questionnaire, but many didn’t. Of the ones who did, many declined subsequent followups. By the time we get to the most detailed tests, the investigators are down to hundreds rather than thousands of data points. Nonetheless, they extrapolate from this self-selected group to the entire population. If we assume that parents who had some concerns about their children were more likely to follow up with the study – a reasonable assumption – then the extrapolation fails.

I’m not saying the results are completely bogus. The researchers are aware of their tools’ limitations, and they take efforts to control for some of them. Still, we have to ask whether a child who’s enrolled in regular school programs, hasn’t been identified as sick by any of his or her teachers, and seems to be progressing just fine in life needs to be given a diagnosis. Maybe there are a lot of children who aren’t truly “normal,” but who are acting the part well enough to pass. On some level, doesn’t that describe us all?

Brugha, T., McManus, S., Bankart, J., Scott, F., Purdon, S., Smith, J., Bebbington, P., Jenkins, R., & Meltzer, H. (2011). Epidemiology of Autism Spectrum Disorders in Adults in the Community in England Archives of General Psychiatry, 68 (5), 459-465 DOI: 10.1001/archgenpsychiatry.2011.38

Kim, Y., Leventhal, B., Koh, Y., Fombonne, E., Laska, E., Lim, E., Cheon, K., Kim, S., Kim, Y., Lee, H., Song, D., & Grinker, R. (2011). Prevalence of Autism Spectrum Disorders in a Total Population Sample American Journal of Psychiatry DOI: 10.1176/appi.ajp.2011.10101532