We live in a world of Big Data, spending much time searching for meaning in its vast shadows and often emerging from those shadows drawing wildly different conclusions.
But the hard facts of Big Data become so soft and malleable in varied human hands, and solid data points often transform into flexible storylines.
Two weeks ago, another vast shadow of Big Data drifted over U.S. media outlets when the Centers for Disease Control and Prevention reported its upwardly revised prevalence of autism spectrum disorders, or ASD. Its 2010 surveillance of 8-year-old children identified 1 in 68 -- 1 in 42 boys, and 1 in 189 girls -- as having ASD.
Emerging from that report, media headlines often concluded that autism rates were "surging." Others heralded the data as proof of an incremental jump in the number of children living with autism.
But what is true? Was this an increase in incidences of autism, that is to say more children were being born each year with ASD? Or was the prevalence increasing -- meaning only that more children were being identified as having autism?
Regardless of interpretation, that 1 in 68 children has ASD at any age is a matter of great public health concern. It should encourage us to develop a national strategy that not only seeks cause and cure, but also supports the clinical, educational and psychosocial needs of so many affected children and their families.
Still, we want to spend our limited ASD allocations in the most productive and efficient manner. After all, real people needing help are waiting on the research and support.
So, when we make those allocation decisions, it's important to understand whether CDC's Big Data are telling us there has been a rise in the "incidence" or the "prevalence" of ASD.
The distinctions are meaningful and important because they lead us to ask different questions.
Dr. Young-Shin Kim is a world-renowned authority on autism disorders, a psychiatrist-researcher who this month leaves Yale University to join the faculty at UC San Francisco. Her expertise has earned her a Presidential Early Career Award for Scientists and Engineers and multiple academic titles. Understandably, I was happy to have an opportunity to speak with her about the CDC's recent report.
"I was not surprised by the increase in the reported prevalence," she said. "But much of that was due to the methods used to identify and diagnose children -- not necessarily to actual increases in the number of children being born with ASD."
She explained that a major reason for the reported rise in ASD prevalence could be explained by changing diagnostic practices, which now identify more children with milder symptoms in the official count.
"But," she cautioned, "to test the hypothesis of whether autism is actually increasing, we would need to take a look at the incidence of autism, not the prevalence. And since we don't yet have studies of that nature, we really can't say."
So how can we determine whether there is a genuine rise in new cases of ASD? Kim said a longer view of autism over time is needed -- not in one snapshot, as is pictured by "prevalence" data. For each of several years, we'd need to document how many of the children who are born each year are diagnosed with autism over the subsequent decade or so. If that number rises, then we know that the yearly incidence of autism is actually increasing.
So what can we conclude about ASD from this report? Even here, uncertainty arises because the reported prevalence varies greatly, depending upon a child's sex, geographical location or ethnicity.
But, according to Kim, many of those observed variations were likely due to variation in the methods used to identify and count children as having autism.
For example, Kim notes that the CDC data were drawn from a study of 8-year-olds who in 2010 were living at 11 separate sites across the country (not including California). But not all sites offered the same quantity or quality of records to researchers, making it more or less likely for them to find an ASD diagnosis. At several sites the educational records were not accessible, depriving researchers of broader diagnostic perspectives from teachers, other school personnel and parents.
And in medical records, documentation of ASD diagnoses also varied when based upon insurance billing codes -- some systems using broader and more inclusive codes that captured more cases.
Finally, we know that the mere existence of any medical record or billing code to document ASD relies upon another critical variable: whether a child has access to medical care. And that a child is more likely to be seen and "counted" if they live in a community that promotes autism awareness, facilitates diagnostic evaluation and offers supportive services.
At the proverbial end of the day, the CDC report reminds us that statistics are always beholden to a limited set of optics that can render children in need of care to be more or less visible.
In the midst of all the statistical uncertainties, Kim is certain of this advice to parents: "You know your children best. If you feel they are not doing well, if you have any question about their ability to meet their full potential to become happy and productive members of society -- ask a pediatrician or specialist. And if you're not satisfied with the answers, keep asking until you are satisfied."
Kate Scannell is a Bay Area physician and the author, most recently, of the novel "Flood Stage."