A new look at autism’s rise

MIT Sloan PhD candidate digs into dynamics behind diagnosis

June 6, 2014

Phech Colatat, PhD ’14

Phech Colatat, PhD ’14

The headlines are as unavoidable as they are alarming. This year, the Centers for Disease Control and Prevention reported that 1 in 68 8-year-old children were diagnosed with some form of autism in 2010, up from 1 in 150 a decade before. No one knows exactly why that rate keeps rising. Researchers have delved into everything from mercury exposure to parental age. But Phech Colatat, who earns his PhD in economic sociology from MIT Sloan this month, thinks we’ve been looking for answers in the wrong places.

“At the end of the day, it’s pediatricians, psychologists and psychiatrists that are formally making the call” to diagnose a child, Colatat says, and no one has studied that group. Colatat did, and found something surprising: diagnostic practices are heavily influenced by key individuals who establish an approach that then persists over time. His paper “Imprinting variation: The diagnosis of autism spectrum disorder at two specialty clinics” will be submitted for review by the American Journal of Sociology. The paper took second prize—and $2,000—at the second annual MIT Sloan doctoral research forum this spring.

When Colatat took the general exams for his PhD, one of the essay questions was about autism research conducted by Columbia University sociologists. The research had used the largest data set of children with autism—records on tens of thousands of children from the state of California. Colatat passed the exams and began to work on a thesis about the Defense Advanced Research Projects Agency. But that massive trove of autism data had stuck in his mind. He decided to find out why so much remains unclear about autism spectrum diagnosis. He theorized that doctors were the missing piece of the puzzle.

One of his advisors, Professor Ezra Zuckerman, encouraged him to search for data from health care providers. Colatat courted a major California HMO, pitching his project and asking for access and cooperation. It took four months, but it worked—the HMO opened its records on 276,000 patients, 3,000 of whom had been given an autism diagnosis. Colatat’s next challenge was funding. He spent an entire year applying for grants. His third attempt succeeded; the project finally moved forward with support from the Simons Center for the Social Brain at MIT.

The HMO’s data included records from three clinics established specifically for autism spectrum diagnosis. Colatat looked at the chances that a child visiting a clinic would go home with a diagnosis. At two of the clinics, doctors diagnosed 35 to 39 percent of the patients with an autism spectrum disorder. At the third clinic, the rate was 65 percent. Those rates persisted over time, even when Colatat filtered for race, environmental factors, and parents’ education. “The difference was still 30 percent,” he says. Perhaps the most telling discovery: when doctors moved from one clinic to another, their rates of diagnosis immediately changed to match that of the clinic as a whole.

Colatat got permission to embed in the clinics, sitting with doctors and families, sometimes in the exam room and sometimes behind a one-way mirror. He conducted dozens of interviews with the clinicians to get a sense of how they had learned to diagnose autism. What he heard was the same few names again and again. At one clinic, a consultant from a nearby university had served as an intellectual mentor to the staff. She had impressed upon them how subtle the signs of autism can be, and as a result, they tended to give out the diagnosis more readily. At another, the clinic’s first director instilled the belief that autism can look like a lot of other conditions, which caused staff to be more conservative. These charismatic individuals made an impression that lasted.

“They were all trying to do their best job,” Colatat says. “But there’s still a difference, and that reflects the field as a whole.”

The diagnosis of autism spectrum disorders is especially vulnerable to this phenomenon, known as imprinting. That’s because it’s a subjective enterprise; there is no blood test or genetic marker for the disease. But imprinting variation could factor into all kinds of medical diagnoses. In fact, it can account for the ways that any group or company reacts to trends in industry. “It affects how an organization is structured, who they hire, and how they coordinate work,” Colatat says.

MIT anthropology professor Susan Silbey, one of Colatat’s dissertation advisors, says the autism research is sociologically significant because “it provides evidence that even well-trained professionals are affected by the local organization of their work. So much so that expert knowledge can be significantly reshaped by local organizational cultures.”

“Phech has found evidence of a basic social force, and has also found a cause of its variation,” Silbey says.

How do we overcome the variation and get at the truth? Which children should and should not be diagnosed? Colatat has some suggestions.

“A good start would be to rotate staff through different clinics, to videotape and watch each other,” he says. And his work isn’t done. Colatat plans to follow the children and doctors going forward. He’s working on two more papers about how information moves among parent groups and pediatricians, making autism diagnosis more or less likely. But ultimately, Colatat points out, it’s treatment that matters most. “Diagnosis is only one part of the puzzle,” he says. “I’m going to look at how organizational dynamics affect not just diagnosis, but also treatment and developmental outcomes. I’m really excited about the idea that multiple research insights can come out of this project.”