From Forbes Magazine's Pharma & Healthcare Blog
By Emily Willingham
November 13, 2015
Is it vaccines? Air pollution? Infections?
Nope. The reason that the latest numbers for autism prevalence among US children have climbed traces largely to a simple change in how interviewers asked a question.
The US Centers for Disease Control and Prevention last conducted the National Health Interview Survey (NHIS) for the years 2011 through 2013. The 2014 survey, though, included a tweak, and that tweak is the reason that autism prevalence climbed from 1.25% to 2.24% in 2014.
Not even the most die-hard causation theorist could argue that in a single year or handful of years, something environmental, like vaccines, caused a near-doubling of autism prevalence in children ages 3 to 17 years.
So what underlies the increase?
For the 2011-2013 survey, parents answered a series of three questions. The first asked if their child had intellectual disability. The second asked if their child had any developmental delay. And the third question listed several conditions, from Down syndrome to sickle cell anemia to autism spectrum disorder (ASD), and parents were asked if their child had been diagnosed with any of them.
But 2014 brought some tweaks, and those tweaks made a difference. The intellectual disability question came first again. But the second question directly asked parents if their child had an ASD diagnosis. The third question then asked about any other developmental delay. More than 10,000 parents are interviewed in each year of this survey.
The simple change to emphasize the autism question resulted in the near doubling of prevalence from 2011–2013 to 2014. Underscoring that this increase reflects a shift in how parents responded to the questions, the prevalence of 'other developmental disorders’ dropped in that same time period from 4.84% in 2011-2013 to 3.57% in 2014.
Intellectual disability prevalence remained pretty much the same in the two periods, and the collective prevalence for all three conditions (intellectual disability, ASD, and other developmental disorders) also remained stable.
What a difference a question can make. But that might not have been the sole influence on the results.
Paul Lipkin, director of medical informatics at the Kennedy Krieger Institute and of the institute’s Interactive Autism Network, sees the reordering and rewording of the questions as one factor. The changes place a “higher priority on identifying autism,” he said in an email, but “at the same time, we know that professionals, parents, and the public also are more attuned to ASD and its identification at all ages.”
These latest values bring the results of three national surveys of autism prevalence into alignment. In addition to the NHIS, the U.S. also identifies autism prevalence values from the Autism and Developmental Disabilities Monitoring Network (ADDM) and the National Survey of Children’s Health (NSCH).
The most recent results from the NSCH put autism prevalence at 1 in 50 children. This latest NHIS prevalence of 1 in 45 converges on that finding, and the agreement among the studies strengthens their conclusions.
Lipkin noted that the advantages of the NHIS study are that it doesn’t involve preselecting households for autism or other developmental conditions and samples across ages, unlike ADDM, which focuses on 8-year-olds.
The 2.24% prevalence also is remarkably close to the 2.64% reported in a thorough investigation of autism prevalence in the South Korean population. Lipkin sees that similarity as pointing to the universality of autism. “This is not a function of professional practice, cultural differences in parenting, or differing parental perspectives on their children,” he said.
The report also shows some stabilizing of previous disparities and a closing gap between girls and boys diagnosed with autism. Earlier surveys and some research have suggested that autism is four or five times more common in boys than in girls, but the NHIS survey has it as just under 3 times more common in boys (3.29 vs 1.15 in girls).
Some of the gaps among ethnicities also appear to be narrowing: ASD prevalence among non-Hispanic white children was 2.55 in the NHIS survey and 2.21 among non-Hispanic blacks, but still lower among Hispanic children at 1.49. Prevalence also is lower among uninsured families, possibly reflecting an access issue.
One final result to highlight from this survey: The results are similar for age groups, at 2.34 for children ages 3-10 and 2.13 for children ages 11-17. That similarity also suggests some stability across the years in the true prevalence of this condition, whether a child was born in 1998 or in the 21stcentury.
I am a journalist and biologist. My book, "The Informed Parent", with co-author Tara Haelle, is available for pre-order. Read more about me here and find me (too often) on Twitter.