Culture and Society
PUBLISHED
What you'll discover in this article
- A South Korean study screened 7-year-olds for autism over 12 years and compared the number they found with the number who had received a diagnosis or were otherwise known to health services.
- They found that the actual percentage of autistic children remained stable over the 12 years, only the proportion picked up by health services had increased.
- Dr Maureen Durkin, who wasn't involved in the study, told IFLScience there is "no doubt" perceived increases in numbers of autistic people can be explained by expanding rates of diagnosis.
A study of South Korean children has added to evidence that – counter to statements from US government officials – underlying rates of autism aren't actually on the rise, and instead there has simply been an increase in the number of autistic people receiving a diagnosis.
The work attempted to screen the total population of 7-year-olds in a community near Seoul for autism each year for 12 years, and it showed that while the proportion of autistic children varied from year to year – ranging from 1.9 percent for children born in 2002 to 3.2 percent for those born in 1999 – there was no consistent trend of increase or decrease.
In other words, there was no evidence of changing rates of autism in the population or an autism "epidemic." Instead, the researchers found that the proportion of autistic children known to the healthcare system had simply increased over time while the proportion of those outside the system had decreased.
Autism is genuinely broad.
Dr Young Shin Kim
"Imagine that the total number of autistic children remains the same: At first, many are outside the service system and therefore are not counted. As detection improves, those children move from the previously unidentified group into the identified, service-exposed group," lead author Dr Young Shin Kim at the University of Texas at Austin told IFLScience.
"The recorded prevalence rises even though the total occurrence of autism has not changed. This context is essential when interpreting increases in autism prevalence based on clinical or service records."
An epidemic of discovery
For researchers in the field, this isn't necessarily surprising, as the idea of rising autism rates being an "epidemic of discovery," rather than a result of actual numbers of autistic people going up, has been prominent for some time.
Dr Maureen Durkin, principal investigator of the Wisconsin site of the US Autism and Developmental Disabilities Monitoring (ADDM) Network and a researcher at the University of Wisconsin-Madison – who wasn't involved in the new paper – told IFLScience there is "no doubt" rising levels of documented autism in recent years can be explained by expanding rates of diagnosis.
She also points out that all population-based studies of autism have their limitations and weaknesses, but this new work is nevertheless "an important contribution to the literature."
To get their results, Kim's team initially had parents of children at over 70 elementary schools in a satellite city near Seoul complete the 27-item Autism Spectrum Screening Questionnaire. This was done each year between 2005 and 2017 (except for 2011, due to a lapse in funding).
Across the 12-year period over 60,000 children completed screening and, based on their results, 2,077 were invited to receive a comprehensive diagnosis. The final diagnosis was made according to the definition of autism in the Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM-5), published by the American Psychiatric Association.
A broad spectrum
In addition, the researchers recorded details like the sex of the child, whether their autism had been picked up by health services, and what autistic characteristics they displayed.
[Service shortages] should not be addressed by leaving some clinically affected individuals unidentified.
Dr Young Shin Kim
As a result, they conclude that children determined to be autistic during the study who hadn't been picked up by health services tended to have a "milder" presentation and also were more likely to be female than those who had been previously diagnosed.
This is consistent with the idea that more people are autistic than are typically identified by health services, and those people skew more female and have less-apparent autistic characteristics.
"Autism is genuinely broad. Some autistic individuals have profound intellectual, language, behavioral, or medical challenges and need intensive lifelong support. Others have stronger cognitive and language abilities but still experience meaningful difficulties in social communication, emotional health, daily functioning, education, or employment," said Kim.
This breadth of autism presentations has been the subject of recent debate in the autism research space, with some researchers – notably Dame Uta Frith at University College London – arguing that the definition of autism, especially in recent editions of the DSM, has become so broad that the term is "losing its meaning." Frith worries this is resulting in overdiagnosis and diverting support away from people who experience more difficulties due to their autism.
Echoing others who have pushed back against Frith's claims, Kim told IFLScience that any shortage of services was a problem of funding and systems, not one of overdiagnosis, and that "less impaired" people are still meaningfully affected by their autism.
"[The issue] should not be addressed by leaving some clinically affected individuals unidentified," she said.
Kim also cautioned that because the study was conducted in one South Korean community, we can't assume the exact trends identified here hold true in every country or that every reported increase in autism is entirely due to improved detection.
However, the wider lesson is broadly relevant, she said: "Autism prevalence is strongly influenced by how thoroughly we look for it, who has access to evaluation, and whether the same methods are used consistently over time."
The researchers also acknowledge several other limitations. For example, despite their best efforts, participation in screening wasn't ubiquitous and varied between 60 percent and 93 percent depending on the year. In addition, fewer than half of the children who were picked up by screening went on to complete a full diagnostic assessment, a fact that could have introduced bias into the results.
The researchers tried to mitigate this using machine learning to predict what diagnostic results the children who dropped out might have received. While this isn't a substitute for in-person clinical assessments, Kim said they tried a number of different variations of this approach, and the central finding that autism rates remained stable did not change.
The study is published in the journal JAMA Pediatrics.
