This week, the journal Pediatrics released new statistics compiled by the CDC on the prevalence of autism, boosting the rate from 1 in 150 to 1 in 100. It’s another staggering leap in an apparent epidemic—more than doubling the rate of children diagnosed with autism since 1996. Indeed, over the past dozen years, autism has made sad, steady progress from obscure syndrome to seemingly ubiquitous developmental disorder.

In every state, our scary autism epidemic fuels walkathons, awareness events, and a proliferation of local chapters of national autism organizations. And across the country, concerned parents whose children aren’t keeping up or seem troublingly different, turn to medical professionals and early childhood educators for evaluation and help. The problems are real.

But what if the autism statistics are wrong?

In 1987, the Diagnostic and Statistic Manual of Mental Disorders (DSM) began broadening the definition of autism to include not only children for whom socialization is impossible, but also those with varying levels of ability to interact and function. What was once a devastating affliction known simply as “autism” evolved into a “spectrum” of disorders, encompassing everything from profound impairment to far milder challenges. Naturally, this more expansive definition of autism partly explains the exponential increase in diagnoses in recent years.

But many children whose symptoms significantly differ from classic autism—who belong only on the milder end of the autism spectrum, if they belong anywhere on the spectrum at all—are inaccurately ending up with serious autism diagnoses.