All scientific studies tell a story. Sometimes the story is very clear and sometimes hard to find.

A new article on the epidemiology of autism in JAMA, “Autism Diagnosis in Children and Adults in the United States, 2011-2022” has received significant media coverage (The United States today title: “Autism increases in young adults, study finds” for focus on autistic adults.

However, behind the headlines, it’s harder to understand exactly what this study is telling us. I argue that a closer look at the data shows little evidence of a wave of new autism diagnoses among adults – or that there are large numbers of cases. autistic adults who have not yet been diagnosed.

Lead author Luke Grosvenor said The United States today that “researchers involved in the study set out to assess how many adults had been diagnosed with autism, because previous studies focused largely on school-age children.” The document itself states: “The largest relative increase in diagnosis rate between 2011 and 2022 occurred among those aged 26 to 34 (450%). »

What exactly does this mean? You could interpret this to mean that the increase in adults means that health systems have identified and diagnosed a larger portion of existing autistic adults. If so, you’d be in good company. Andy Shih, scientific director of Autism Speaks, wrote on LinkedIn that this study is relevant “especially for adults who may not have been diagnosed.” childhood.”

The study doesn’t say anything about previously undiagnosed adults, but that seems like a possible interpretation. Another interpretation is that the number of autistic adults increased naturally as the number of autistic children grew. These two stories have very different real-world implications.

The story of “recently diagnosed adults” means that there has been little or no real increase in the number of autistic children over time. Simply put, there are a large number of undiagnosed adults and we need to identify them and provide them with appropriate services.

In contrast, according to the “growing number of autistic children” story, the onset rate of autism has increased significantly over the years. This implies that something must have caused this increase, and it cannot simply be inherited. This also means that there are far fewer undiagnosed adults than some people claim.

Obviously these two stories can’t both be correct. And which one is right has huge implications for public policy and our autism research priorities. How do you know which one is true?

The key to understanding the epidemiology of autism is to look at data by year of birth. Your year of birth is one of the few things that can never change about you. On the other hand, your age always increases by one year each year.

When you compare people of a certain age in different years of education, you are comparing different people born in different years. Many articles present data organized by year of study, while few explicitly state year of birth. You can easily convert one to the other: year of study minus age equals year of birth. For example, children who were eight years old in 2020 were born in 2012. JAMA the data reported by the newspaper is in terms of age and not year of birth. But we can easily convert it to year of birth.

To understand what the data tells us, we need to know what the numbers mean. Is it diagnosis rates? Or is it about prevalence, that is, the number of autistic people at any given time? They are different things. The document uses both “diagnosis rates” and “prevalence.”

Although the correct interpretation is not obvious, Grosvenor confirmed my guess. In fact, the data shows the proportions of participating health network members who had an autism or ASD diagnosis code in their electronic health record during each study year. A diagnosis code does not mean that someone has been recently diagnosed. This does not necessarily mean they have already been diagnosed. Physicians treating patients decide which diagnosis codes to enter in patient records. An autism or ASD diagnosis code may indicate that the patient was diagnosed at some point in the past, but this is not completely accurate. We can treat it as prevalence.

The document includes tables showing some of its data. This data is sufficient to compare the two conflicting stories: “recently diagnosed adults” versus “growing number of autistic children.”

Is it possible that the newspaper’s report of a sharp increase among adults aged 26 to 34 shows that the “newly diagnosed adults” story is true? No. Why not ? Because it compares two different groups of people, born at different ages.

The 26-34 year old group in the 2011 data was born between 1976 and 1984, while the 26-34 year old group in the 2022 data was born between 1987 and 1995. (In the data in this article, ages refer to the year before each year of study ) A higher diagnosed prevalence in the later-born group than in the earlier-born group could simply be due to a higher prevalence among this group.

But does the group of children born later really have a higher prevalence? We can answer this question by taking a closer look at the data.

The diary data contains information on prevalence by year of birth, but it takes a bit of effort to extract it. This JAMA Electronic Table 7 in the article gives the proportions of autistic patients (those with diagnostic codes) for each year of study grouped into age groups. I used an Excel spreadsheet to study the prevalence of diagnosis codes by year of birth. The result is not as precise as we would like, due to the age range, but it is still quite informative.

The following graph shows prevalence by birth year for birth years 1976 to 1995, encompassing both the earlier and later groups mentioned above. As I said, it’s not exactly prevalence, but it’s the closest we can get to it with this data.

As you can see, there was a dramatic exponential increase in birth year prevalence: 12.3% per year. Note that the chart appears lumpy because the data is grouped by year ranges.

The group of children born later, born between 1987 and 1995, has an average prevalence during the year of birth which is 3.82 times higher than that of the previous group born between 1976 and 1984. Thus, the majority, or all of the increase observed between these two groups of adults probably came from the increase in prevalence during the year of birth. This shows that the interpretation of “recently diagnosed adults” is not correct.

Year of birth Prevalence 1976 – 1995 according to Grosvenor et al.

Source: Alexander MacInnis

Other data sources consistently show that as people in each birth year get older, more people are diagnosed, and this trend is similar across all birth years. So even if health systems were able to diagnose some members as adults – we can’t tell from this data – it’s not something new. This has been going on for decades.

We can also examine the prevalence trend by year of birth for all years of birth. Here is a chart for birth years from 1990 to 2013, the last birth year that was 8 years old at the time of the last study year.

Year of birth Prevalence 1990 – 2013 according to Grosvenor et al.

Source: Alexander MacInnis

That’s a 7 percent increase per year of birth, which seems pretty significant to me. That’s almost the same increase as the CDC’s autism reports overall for children born 8 years before each study year. CDC reports show an increase of 7.4 percent per year between birth years 1992 and 2012.

Two scientific papers published in recent years purport to show that there are large numbers of undiagnosed adults. Both rely on the assumption that prevalence by year of birth has remained constant from the beginning.

I submit that the data contained in this document JAMA The article and the CDC reports contradict this hypothesis, meaning that there are relatively few undiagnosed adults today, but there will be many more autistic adults in the future as young people with autism are getting older. This says a lot about where public policy and research should focus.