Why Changing Autism Numbers Do Not Equal ‘Epidemic’

Last week Dr. Bennett Leventhal, a psychiatrist whose considerable expertise includes the epidemiology of psychiatric illness, stopped by our office to give an overview of the state of autism spectrum disorders. Dr. Leventhal gave a wide-ranging talk on everything from causes to treatments to what’s really going on with the so-called “epidemic” of new cases.

“Has the prevalence been rising?” Dr. Leventhal asked. “For sure it’s been rising. When I was in training it was 4 per 10,000 live births. That’s 0.04 percent. It’s been going up and up and up, and in 2011 we published what I still think is the best paper on autism prevalence, which said the rate is 2.64 percent.” But understanding why it’s happening—and why it isn’t—is a manner of soberly assessing the available information and make sure that everyone is “counting the right way.”

Dr. Leventhal described the development of a standardized diagnosis in the 1970s, thanks to the advocates of the National Society of Autistic Children. In 1978, the organization developed criteria that would soon contribute to the autism diagnosis in the DSM-III, of 1980. “It’s important to know how we got here,” Dr. Leventhal said, “and it wasn’t a bunch of smart guys sitting around—it was a bunch of moms who said, ‘You scientists need to do things differently, you clinicians have to do things differently.’”

A key factor that contributed to the consistent increases in prevalence that continue to this day is education law. “This prevalence started rising in the 1970s,” he said. “In 1975 Congress passed Public Law 94-142, the Education for All Handicapped Children Act, which said you can’t throw kids out of school because they have a disability. And by the early 1990s the DSM-IV came out and those became the accepted criteria in schools.The vast majority of these increases in prevalence come from service data.”

But there is another lesson to be had here. “The real critical point to remember is that there is a difference between prevalence and incidence. Prevalence is the number of people in the population who have a condition. And incidence is the number of new cases coming into the population. Increases in incidence represent epidemics—all of a sudden a lot of people have measles. But can you increase prevalence without having any more new cases? Absolutely, and it’s really simple.

“How do you do that? The first thing is you can change the diagnostic categories,” which have been getting consistently broader, said Dr. Leventhal. “You can change the way you do diagnosis, with better tools for diagnosis. You can increase awareness. And you can also find out that there were methods problems” in the past that artificially kept the prevalence down.

“The usual diagnosis of autism in the 1970s began at the age of 6 or 7 when a kid showed up in school. And then they stopped having autism at 16. Why? Because they left school and were sent to institutions or other places where they were called mentally retarded. So if you’re only counting people between 6 and 16, the denominator stays the same and the numerator is small”—that is, the population at large is constant but the subset of people with autism is restricted to a strict age range. But when we understand that autism is a lifelong disorder, things change. “If you all of a sudden add kids from 2 to 6, the numerator goes up but the denominator remains the same.” And if you add everyone with autism over the age of 16, “Voila,” said Dr. Leventhal. “Your prevalence has risen without changing the number of people with the disorder. You’re just counting everybody, and you’re counting the right way.”

For instance, the study Dr. Leventhal mentioned above, which returned a staggering prevalence of 2.64 percent, was what’s called a full population study of Korean youth, endeavoring not to just get a representative group but to test everyone in a community. By “counting everybody,” the study was able to identify a “non-clinical” population—kids who didn’t have autism diagnoses, and weren’t getting services—who in fact met criteria for the disorder.

Dr. Leventhal is skeptical of the concept of an autism epidemic for a similar epidemiological rationale, and the culprits are usually methods problems. These plague every area of causal inquiry, he said, from genetics to environmental to social. Counting the right way may give us a higher prevalence—but counting the wrong way creates false hopes and false fears.

Dr. Leventhal is co-chair of the Child Mind Institute’s Scientific Research Council, the Irving B. Harris Professor of Child and Adolescent Psychiatry, Emeritus, at the University of Chicago, and deputy director of the Nathan S. Kline Institute for Psychiatric Research.