The question of the day: Is there an autism epidemic?

The answer? We don't really know, according to Craig Newschaffer, PhD, associate professor at the Johns Hopkins School of Public Health. But fears of an autism epidemic have been overblown by the press and advocacy groups, said Dr. Newschaffer, speaking at the American Academy of Pediatrics National Conference and Exhibition today in Washington, D.C.

Dr. Schaffer, director of the Center for Autism and Developmental Disabilities Epidemiology at Hopkins, said that it's impossible to know whether autism is simply being diagnosed more often, or whether any factors, such as environmental issues, are playing a role. He noted that the criteria for diagnosis of autism has changed over time and that professionals may be interpreting the wording in the diagnostic definition differently.

He also noted that research widely reported in the media has not been published in a peer review journal. As one example of how things have played in the press, Dr. Newschaffer pointed to a highly publicized 2002 report to the California legislature that said the proportion of mental retardation cases meeting autism criteria had not changed between two birth cohorts separated by a 10-year interval. But of the two birth cohorts, the second (from 1993-1997) was shown to have less than half the rate of intellectual impairment of the earlier one, suggesting that the cohorts studied really were not similar.

So how do we find out if autism is occurring more often? Dr. Newschaffer offered a number of recommendations:

• Carry out research on trends in diagnostic practice over time
• Bring in social science to look at the phenomenon of trends in diagnosis
• Conduct more simulation studies with the data, showing, for example, what happens with the statistics when various measures are changed by differing amounts

Dr. Newschaffer believes that answers to questions about the prevalence of autism may be coming in the next five to 10 years. The public health surveillance system now in place will much more rigorously define when a child is autistic. In the meantime, he urged physicians and others to stay away from what he called three spurious arguments: that an apparent increase in prevalence reflected in studies is so large that it must be real; that potential autism risk factors of "of x, y, and z" have increased in prevalence and, therefore, the autism increase must be real; and that all of such an apparent increase must be the result of diagnostic change because autism is genetically determined.