Autism: What's in a Number?

Feb. 9, 2007 — -- It's all over the wires, the papers and the airwaves: A new report from the Centers for Disease Control and Prevention finds that autism is more prevalent than ever before estimated, affecting roughly 1 in 150 American children.

USA Today led with that statistic today, as did "World News" last night -- both standing bravely in the face of the news hurricane called Anna Nicole Smith. The Washington Post has it on page A6, The New York Times on page A12: "Study Puts Rate of Autism at 1 in 150 U.S. children."

But does it?

According to the CDC, it does not.

But before we blame the media, we might take a look at the source of the confusion -- the news conference the CDC held to announce the study Thursday.

Its experts came awfully close to having it both ways, at once repeatedly cautioning against using the 1 in 150 number as a national estimate, and then suggesting that, well, you could sort of tentatively do just that -- to the point of calculating that 1 in 150 would extrapolate to 560,000 American children with autism (specifically, with autism spectrum disorders, or ASDs).

By contrast, the study itself and the CDC's printed materials announcing it don't go there at all. They explicitly say the study's results are not nationally representative. And the study cites earlier estimates of autism prevalence that -- while not strictly comparable -- are higher than its own.

"We're not generalizing these results to the entire United States," Marshalyn Yeargin-Allsopp, chief of the CDC's autism program, told me this morning. As for the "highest ever" claims, she said, "They're not higher than any previous estimates."

Compare that to yesterday's Reuters copy: "Autism is more common in the United States than anyone has estimated, affecting about 1 in every 150 children, the U.S. Centers for Disease Control and Prevention reported on Thursday."

Suffice it to say that what we have here, à la Cool Hand Luke, is a failure to communicate.

Findings Not Nationally Representative

The true font of available wisdom on the study is the study itself, a 12-page document with a 26-word title published in this week's issue of the CDC's Morbidity and Mortality Weekly Report (the same publication that on June 5, 1981, carried the very first report of the disease that later became known as AIDS).

Can the autism study's figures be used to produce a national estimate of ASDs? The report is unequivocal: "Because participating sites were selected through a competitive federal award process and not specifically to be representative of the entire U.S. population, ADDM Network results cannot be used as a basis for estimating the national prevalence of ASDs."

In the very next paragraph, the report effectively knocks down the notion that this estimate is higher than any previously made. It notes that two earlier studies -- both designed to be nationally representative -- produced estimates of ASDs among 6- to 8-year-olds of 7.5 and 7.6 per 1,000. That is 1 in 132, which is more than 1 in 150.

All these studies have limitations, and many produce a range of estimates. The new one -- done among 8-year-olds in 14 states -- found as few as 1 in 303 cases in Alabama and as many as 1 in 94 in New Jersey.

The CDC in the past has used various data sources to cite a range from 1 in 500 to 1 in 166; the latter number is the best known because it's been cited in public service advertisements by Autism Speaks, an autism awareness group.

The difference between 1 in 166 and 1 in 150 apparently sparked the reporting about a new, higher CDC estimate of autism prevalence, even though that change is so slight as to lack any real meaning, especially given the lack of comparability of the source data.

Instead, says Yeargin-Allsopp, when the CDC says the new estimate is higher than in the past, it's referring to studies with much lower estimates that were done in the 1990s, not to more recent ones.

ASDs Still Critical Public Health Issue

None of this takes away from the importance of ASDs, their enormous impact on affected families and the obvious need for good data, diagnosis and medical research into their causes. Indeed, the importance of the issue makes the need for good information, accurately reported, all the more critical.

Did the CDC and the media's reporting of its numbers move us in that direction yesterday? Look at the excerpt from the news conference below -- and the coverage that followed -- and make your own call.

WILL DUNHAM, REPORTER, REUTERS: Yes, hello. I know that you have not given a nationwide rate, but is there any way you could give a ballpark estimate of how many millions of children nationwide you think fall under this category?

DR. YEARGIN-ALLSOPP: Yes. You're absolutely correct. We have not given a national statistic, and we have given a lot of caveats around the 1 in 150 children. However, if there are 4 million children born in the United States every year, which is the current estimate, if we apply these statistics to the population of children from birth to 21, then we get a number, of 560,000 children in the United States with an ASD.