CDC Raises Autism Estimate

It's one of the more contentious issues among scientists, advocates for children, parents and pediatricians. What causes autism? There hasn't even been agreement about how many children in the United States have autism or autism spectrum disorders (ASD). Using data from several studies, the Centers for Disease Control and Prevention determined about five years ago that the rates were probably somewhere between 1-in-500 and 1-in-166 children. Today, the CDC announced the results of its first national data collection on ASD. It concluded that the U.S. rate among 8-year-olds is approximately 1 in 150. NEWSWEEK's Julie Scelfo spoke with Dr. Marshalyn Yeargin-Allsopp, chief of the CDC's developmental-disabilities program at the National Center on Birth Defects and Developmental Disabilities about the study. Excerpts:

NEWSWEEK: One in 150 is higher than the 1-in-166 estimate the CDC has used for the past few years. Are you saying autism rates have gone up?
Marshalyn Yeargin-Allsopp: No. For several years on the CDC Web site we have said the rate of autism is 2 to 6 per 1,000. That number was determined from a review of all the existing studies using different methods all around the world. The upper range was 6 per 1,000 or 1 in 166. The importance of the results published today is that we are confirming the prevalence in the U.S. is about 1 in 150, or at least that's the figure from 14 communities we were able to study.

What communities did you study?
We gathered data in Alabama, Arizona, Arkansas, Colorado, Georgia, Missouri, Utah, West Virginia, Maryland, North Carolina, South Carolina, New Jersey, Pennsylvania and Wisconsin. In some cases the research came from several counties and in other cases it covered the entire state. Participating researchers are part of the CDC's Autism and Developmental Disabilities Monitoring Network, a program we began in 2000 for which participants were carefully chosen.

So what you're saying is not that the rates of autism have gone up, just that now we have some more definitive data?
Precisely. We have more accurate data across multiple sites in the U.S. that allows us to say with more certainty what the prevalence is. This establishes a baseline for us so moving forward, we can better determine whether these rates are increasing, decreasing or remaining stable over time.

How did the researchers gather data?
We had requirement that researchers study an area encompassing a certain number of births. Then they gathered data from multiple sources include psychologists, pediatricians, hospital clinics serving children, as well as state run programs for children with developmental disabilities. In addition, researchers in some states were able to gather data from school systems.

Were there any regional variations?
The consistency across states was surprising to us. It validated the methods of our data collection. There were, however, two states that fell outside the surveillance site average. In New Jersey, the rate is 10.6 per 1,000. And in Alabama, the rate is 3.3 per 1,000.

Why are the rates so different in those two states?
In Alabama, the researchers were unable to get access to the education records of children in the counties they studied. And in New Jersey, there was a great deal of information available in the children's records, which made it easier for researchers to identify ASD.

Are you saying that if other states had as detailed case files as New Jersey that the rates would likely be higher?
That is possible.

Why does New Jersey have such thorough records?
New Jersey has a long history of providing services for individuals with developmental disabilities. So there may be a better history of documentation of developmental disabilities than there is in other states.

How does the 1-in-150 figure compare to other countries?
Interestingly, there have been two studies recently outside the U.S. that have reported prevalence rates as high as in New Jersey: Sweden and the U.K. We will continue measuring prevalence rates every two years at least until 2010, and we will also be looking at the quality and information available in children's health and education record in order to understand how that affects prevalence estimates.