Are autism rates really rising?
Not in this study. When researchers screened South Korean children the same way for 12 birth years in a row, the share with autism stayed flat. Autism rates by age 7 held between 1.9% and 3.2%, with no statistically significant rise or fall.
Reported autism rates have climbed sharply since the 1980s, and that climb has worried a lot of parents. But most of those numbers come from records of diagnoses and services. The authors point out that records like these can shift when diagnostic habits change, when more families seek services, and when policies reward a diagnosis. A count can rise even if the true number of children with autism stays the same.
This study, published in JAMA Pediatrics, took a different approach. Instead of counting diagnoses already on file, the team went looking for autism directly, using the same method year after year.
How the autism screening worked
The researchers followed 12 consecutive groups of children born between January 1998 and December 2010. Each group was checked at age 7, between 2005 and 2017. Every child entering elementary school at that age was part of the target group, more than 62,000 children in all. Every year, the same standard screening tool was used: the Autism Spectrum Screening Questionnaire.
Of those children, 2,077 screened positive and were invited for a detailed evaluation, and 881 completed it. That evaluation was thorough. It included two standard autism assessments (the Autism Diagnostic Observation Schedule and the Autism Diagnostic Interview-Revised), cognitive testing, and a review of records. Clinicians then agreed on a best-estimate diagnosis using the current psychiatric diagnostic manual, the DSM-5. The team also noted whether each child had ever been diagnosed or treated before.
What the data show on autism rates
The study reported two closely related measures. One is how many children had developed autism by age 7. The other is how many had autism at age 7.
By age 7, between 1.9% and 3.2% of children had developed autism, depending on birth year. That is roughly 19 to 32 of every 1,000 children. The share with autism at age 7 ranged from 2.0% to 3.4%, or about 20 to 34 per 1,000. The lowest yearly figure was very likely somewhere between 1.5% and 2.4%, and the highest between 2.7% and 3.8% (95% confidence intervals). The numbers moved a little from year to year, but across all 12 birth years there was no statistically significant increase or decrease.
The second finding concerns girls. Among children with autism who had never been diagnosed and were found in regular classrooms, girls made up a disproportionate share. They showed distinct patterns of traits, and many had substantial clinical needs that were going unmet.
Dr. Kumar’s Take
This study answers a question most autism statistics cannot. A diagnosis count mixes two things: how many children have autism, and how hard the system is looking for them. When awareness grows, when definitions change, and when a diagnosis opens the door to services, the count climbs. None of that requires more children to have autism.
By screening every child with the same tool every year, this team took the counting system out of the equation. The result was flat. The authors conclude that detection may account for a large part of the reported increase. That does not rule out that some factors raise or lower autism risk. It does make a large, hidden surge in true cases hard to support, at least among these children.
The finding about girls matters just as much. These were girls sitting in regular classrooms with autism and real needs that no one was meeting. A system that finds children only when someone refers them will keep missing them.
Limits of this autism study
This was one country, and the youngest children in it were born in 2010, so it cannot speak to more recent birth years. South Korea’s schools and culture also differ from those in the US, so the exact rates may not carry over.
Only 881 of the 2,077 children who screened positive completed the full evaluation. When more than half of flagged families do not come in, the children who are missing could shift the estimates. A child the questionnaire missed never reached the full assessment at all. What the study does well is consistency: the same method, applied the same way, across 12 birth years. That consistency is exactly what most autism rate reports lack.
Practical Takeaways
- A rising autism count in the news is not, by itself, proof that more children are developing autism, because counts also rise when screening and diagnosis expand.
- Autism in girls can go unrecognized in regular classrooms, so if you have concerns about your daughter’s social or communication development, ask for an evaluation even if school has not raised a problem.
- A diagnosis at age 7 or later still matters, since the undiagnosed children in this study had substantial needs that were not being met.
- When you read a report on autism rates, check whether it screened every child the same way or simply counted diagnoses in records.
Related Studies and Research
- Brain scans reveal autism may be two different conditions
- Does the MMR vaccine before age 2 cause autism?
- Does birth order affect allergies, ADHD and autism risk?
- Can a baby’s gut bacteria shape autism and ADHD risk?
- Is ADHD linked to constipation and other gut problems?
FAQs
Is autism more common now than in the 1980s?
Reported autism rates are far higher now than in the 1980s. The question is whether that reflects more children with autism or better and broader counting. This study did not reach back to the 1980s, but across 12 birth years it found no real change when the method stayed fixed. Its authors conclude that detection-related factors may explain a large share of the reported rise. Comparing rates across decades is only fair when the same method is used at both ends.
Why do girls with autism go undiagnosed?
In this study, girls made up an outsized share of children with autism who had never been diagnosed and were attending regular classes. Their traits also looked different from the typical pattern. A child who is keeping up in a regular classroom is less likely to be referred for an evaluation. Screening every child, rather than waiting for a referral, is one way to find those who would otherwise be missed.
What is the difference between autism incidence and prevalence?
Incidence counts how many children develop a condition over a period of time, here from birth to age 7. Prevalence counts how many have the condition at a single point, here at age 7. For a lifelong condition like autism, the two numbers tend to be close, and in this study they were. The key point is that both stayed flat across the 12 birth years, so neither measure showed a rise.
Bottom Line
When South Korea’s 7-year-olds were screened with the same questionnaire for 12 birth years in a row, the share with autism held steady at roughly 2% to 3%, with no statistically significant rise. That points to changes in detection and counting, not a surge in children developing autism, as a major driver of the reported increase. The study also found girls with autism in regular classrooms who had never been diagnosed and had needs no one was meeting. The takeaway is less about a rising tide of autism and more about finding the children who are still being missed.
