article

03.30.12

Why New CDC Numbers on Autism May Not Add Up

Shocking new statistics still might not reflect its true prevalence rate. As the DSM gets set to redefine the spectrum, can researchers ever properly calculate the toll of the disorder?

Startling news emerged from the Centers for Disease Control and Prevention Thursday: the agency now estimates that one in every 88 children in America has an autism-spectrum disorder.

The number has been steadily increasing throughout the last decade.

In 2007, the CDC estimated autism prevalence rates at one child in every 150. The new figure represents a 78 percent increase from the one reported 10 years ago.

On the surface, the CDC announcement would seem to be yet another twist in the ever-changing story of autism. The most basic facts of this disorder—what causes it and who has it—are moving targets, subject to perpetual revision and uncertainty.

The news, for instance, follows another much-publicized upheaval in autism research. Early this year, there was widespread concern over proposed changes to the autism diagnosis itself—including the removal of Asperger’s syndrome as a diagnosis—that will take effect next year when the fifth edition of the Diagnostic and Statistical Manual (DSM) of mental health, American psychiatry’s official handbook, is published.

Through all these fractious diagnostic iterations, a fundamental question remains unanswered: is autism occurring more frequently or is it simply being diagnosed where it hadn’t been before?

The new CDC data don’t provide an answer.

The announcement didn’t come as a shock to many researchers versed in the literature of this disorder. 

Pat Levitt, director of the Zilkha Neurogenetic Institute at the University of Southern California, for instance, regarded the new CDC estimates as a “fine tuning” of autism’s demographics, rather than a radical rethink.

“It goes from point 0.85 percent to 1.1 percent [of the total population of American 8-year-olds],” he said.

The CDC data were drawn from an analysis of pre-existing health records from roughly 340,000 children—from 14 sites across the country—all of whom were 8 years old in 2008.

In essence, said Levitt, what the new CDC figures reflect is that there are “a few thousand more children nationally” with autism than previously identified.

The reasons for the expanded prevalence, of course, can’t be accounted for with any single explanation.

But to parse the CDC numbers a bit:

The largest changes in diagnostic status were made among black and Hispanic children.

This is unsurprising. Some studies already suggest that children from minority racial groups in the U.S. are less likely to be accurately diagnosed with autism than are white children. 

The other big change came among children without an intellectual disability—which is to say, those with an IQ above 70. In general, these are the kids who would qualify for a diagnosis of Asperger’s syndrome, commonly understood as high-functioning autism.

These are also the children who are most likely to be affected by the DSM’s changes next year—and who, many believe, stand to lose their diagnosis when Asperger’s leaves the official lexicon. Because of these pending changes to the DSM, it is possible that the new CDC numbers might not hold true a year from now.

Indeed, the CDC estimate is far from definitive.

Self-proclaimed as “not intended to form a nationally representative sample,” the CDC data came from 14 sites—not enough to capture a full picture.

“It’s not a representative sampling of the United States. If you don’t use a representative population, then your findings don’t generalize to the whole population,” says Mike Milham, a neuroscience researcher at the Child Mind Institute in New York. “Especially in the United States,” he added, “where you have such heterogeneity.”

Furthermore, the CDC used a “surveillance method” that reexamined health-related paperwork that had been issued on the children in 2008. The majority of the kids in the sample who've been identified as having ASD by the new CDC estimates had already been given that diagnosis.

So, while the CDC investigation is useful, it’s not a scientific game changer. Both Levitt and Milham pointed to a different piece of research as an example of a truly startling finding. A South Korean study, published last year, showed that one in 38 children in three different suburban areas were on the autism spectrum.

What the CDC study doesn’t necessarily show, says Milham, is an increase in actual cases of autism. Rather, it’s possible that “it’s more about our ability to detect new cases.”

Unlike the CDC report, that investigation took a true epidemiological approach. The researchers looked at all schoolchildren across the three districts, rather than picking out the ones who’ve already been pulled aside by school officials or doctors and reexamining their old paperwork.

Yet capturing a complete picture of the condition in the U.S. is vastly complicated—much more so than in a country like South Korea, where the population is relatively isolated and homogenous.

What the CDC study doesn’t necessarily show, says Milham, is an increase in actual cases of autism. Rather, he says, it’s possible that “it’s more about our ability to detect new cases.”

Of course, Milham says, the importance of detection should not be understated. Just the opposite—early detection is crucial for effective intervention.

“If we’re not picking up on the disorder, then we’re leaving kids untreated,” Milham said. “That’s where the tragedy of it lies.”