Autism prevalence increases, and antivaxers blame it on vaccines (2018 edition)

Every couple of years, the CDC issues its report on the prevalence of autism, which has been rising steadily for the last couple of decades. Somehow, this year I didn’t get to it in a timely fashion, but, as I like to say, better late than never. Besides, every time autism prevalence ticks upward, antivaxers lose it and do their damnedest to blame the increase on vaccines. The same thing happened after the CDC issued a press release a week and a half ago, “Autism prevalence slightly higher in CDC’s ADDM Network“, about its report, Prevalence of Autism Spectrum Disorder Among Children Aged 8 Years — Autism and Developmental Disabilities Monitoring Network, 11 Sites, United States, 2014. Not surprisingly, the antivaccine movement pounced on this report, which showed a modest increase in prevalence since the last CDC report in 2016, as proof that it must be the vaccines, because to antivaxers it’s always—always—about the vaccines. But is it? Of course not. However, as usual, unpacking the data is far more complicated than antivaxers like to make it sound.

The CDC report on autism prevalence – 2018

Before discussing the CDC report’s findings, it’s important to recap what the Autism and Developmental Disabilities Monitoring (ADDM) Network is, as it’s been a few years since I’ve written about its results. In brief, the Children’s Health Act authorized the CDC to monitor the prevalence of autism spectrum disorder (ASD) in multiple areas of the United States, a charge that led to the formation of the ADDM Network in 2000. The CDC notes that since that time it has funded grantees in 16 states, tracks ASD in metropolitan Atlanta, and represents the Georgia site collaborating with competitively funded sites to form the ADDM Network.

The goals of the ADDM are:

• To provide descriptive data on classification and functioning of the population of children with ASD.
• To monitor the prevalence of ASD in different areas of the United States.
• To understand the impact of ASD in U.S. communities.

To accomplish these goals, the CDC set up the ADDM as an active surveillance system that provides estimates of the prevalence of ASD among children aged 8 years whose parents or guardians reside within 11 ADDM sites in the United States (Arizona, Arkansas, Colorado, Georgia, Maryland, Minnesota, Missouri, New Jersey, North Carolina, Tennessee, and Wisconsin), as well as other characteristics, of children aged eight years whose parents live in the ADDM sites. This age was chosen because the baseline ASD prevalence study conducted by the Metropolitan Atlanta Developmental Disabilities Surveillance Program (MADDSP) suggested that this is the age of peak prevalence. The ADDM is the largest population-based program to monitor autism (with more than 325,000 children in the sites under surveillance, or 8% of the US population aged 8 in 2014) and the only autism tracking system that examines health and education records. In other words, in the world of autism, the ADDM is a Big Deal, as are the periodic CDC reports on the data from the network.

Here is how the ADDM ascertains cases:

ADDM is an active surveillance system that does not depend on family or practitioner reporting of an existing ASD diagnosis or classification to determine ASD case status. ADDM staff conduct surveillance to determine case status in a two-phase process. The first phase of ADDM involves review and abstraction of children’s evaluation records from data sources in the community. In the second phase, all abstracted evaluations for each child are compiled in chronological order into a comprehensive record that is reviewed by one or more experienced clinicians to determine the child’s ASD case status. Developmental assessments completed by a wide range of health and education providers are reviewed. Data sources are categorized as either 1) education source type, including evaluations to determine eligibility for special education services or 2) health source type, including diagnostic and developmental assessments from psychologists, neurologists, developmental pediatricians, child psychiatrists, physical therapists, occupational therapists, and speech/language pathologists. Agreements to access records are made at the institutional level in the form of contracts, memoranda, or other formal agreements.

It’s actually far more complex than that. For instance, certain education and health records were missing for certain children, including records that could not be located or whose retrieval would be impractical or too costly; so sensitivity analyses had to be carried out to determine whether these missing records would have an effect on case ascertainment. Another example includes the statistical tests that need to be done to provide confidence intervals and to determine prevalences in different racial groups. A total of 53,120 records for 42,644 children were reviewed from health and education sources. Of these, the records of 10,886 children met the criteria for abstraction. This represented 25.5% of the total number of children whose source records were reviewed and 3.3% of the population under surveillance. Of the records reviewed by clinicians, 5,473 children met the ASD surveillance case definition.

The 2018 report covers the year 2014, which includes the cohort born in 2006. Its key findings include that:

• Autism prevalence has increased to 16.8 per 1,000 (one in 59) children aged 8 years
• Estimated prevalence of ASD was highest in New Jersey (29.3 per 1,000 or 1 in 34.1) compared to each of the other ten sites
• ASD prevalence was 26.6 per 1,000 boys (one in 37.6) and 6.6 per 1,000 girls (one in 151), for a prevalence ratio of 4.0.
• Male-to-female prevalence ratios ranged from 3.2 (Arizona) to 4.9 (Georgia)
• Estimated prevalence among white children (17.2 per 1,000) was 7% greater than that among black children (16.0 per 1,000)
• Estimated prevalence among white children was 22% greater than that among Hispanic children (14.0 per 1,000)
• Estimates for Asian/Pacific Islander children ranged from 7.9 per 1,000 (Colorado) to 19.2 per 1,000 (New Jersey) with wide confidence intervals.

The accompanying press release notes:

The latest estimate of 1.7 percent (1 in 59) is higher than the previous ADDM estimate released in 2016, which found a prevalence of 1.5 percent or 1 in 68 children. Some of the change in prevalence could be due to improved autism identification in minority populations – although autism is still more likely to be identified in white children than in black or Hispanic children. This identification is important, because children identified early with autism and connected to services are more likely to reach their fullest potential.

“Autism prevalence among black and Hispanic children is approaching that of white children,” said Stuart Shapira, M.D., Ph.D., associate director for science at CDC’s National Center on Birth Defects and Developmental Disabilities. “The higher number of black and Hispanic children now being identified with autism could be due to more effective outreach in minority communities and increased efforts to have all children screened for autism so they can get the services they need.”

So, basically, ASD prevalence increased approximately 20% from 2012 to 2014. However, autism identification in minorities like African-Americans and Hispanics is improving, although it is diagnosed more frequently in white children.

There are, of course, several weaknesses with the ADDM estimates, as is the case with all large epidemiological studies. First, the areas that the ADDM covers are not necessarily representative of the rest of the US, and even the areas chosen in each ADDM area are not necessarily representative of the state as a whole. There are also limitations in the medical and educational records in terms of the information available. For instance, some children might have had earlier diagnoses that were not recorded in the records examined. Finally, comparisons with results from earlier ADDM surveillance years were not always restricted to a common geographic area, which led the authors to point out that “inferences about the changing number and characteristics of children with ASD over time should be made with caution” and that “although study methods and geographic areas of coverage have remained generally consistent over time, temporal comparisons are subject to multiple sources of bias and should not be misinterpreted as representing precise measures that control for all sources of bias.” In other words, when the increase in autism prevalence is modest, as it is in this report, one must interpret the data with caution, as there might in fact be significant less (or more) of an increase than is reported.

The bottom line is that, as has been observed in previous reports, this year’s CDC estimate for autism prevalence based on the ADDM has increased, albeit modestly. For instance, in 2012, the estimate of ASD prevalence in the year 2008 was 11.3 per 1,000 (one in 88), and in 2014 the estimate for 2010 was 14.7 per 1,000 (1 in 68). This increase really shouldn’t be too surprising, as it has been widely believed that autism diagnoses have been underreported for a long time. Likely we’re catching up and getting closer to the “true” prevalence of the condition. But what does this all mean? First, I’ll discuss what it doesn’t mean.

Antivaccine groups react

In 2016, as in 2014, as in 2012, as in 2010 (and, I’m sure, but don’t remember) every two years going back to the beginning of the ADDM, when the CDC reported its estimate for autism prevalence, the antivaccine movement tried its very best to blame vaccines. I noticed something different in 2018. For once, nowhere in the press releases from the usual antivaccine suspects were vaccines explicitly mentioned. (Oddly enough, even the first post by the antivaccine blog Age of Autism mentioning the study that didn’t just regurgitate another blog’s post doesn’t mention vaccines, although the commenters immediately did.) For instance, Robert F. Kennedy, Jr.’s World Mercury Project (a group dedicated to fear mongering over mercury in vaccines as a cause of autism and health problems) didn’t mention vaccines once in its post CDC Still Paralyzed by Autism Epidemic: Report Shows One in 59 Children in the US Now Affected:

Yesterday, the Centers for Disease Control (CDC) released the latest autism spectrum disorder (ASD) prevalence estimate from its Autism and Developmental Disabilities (ADDM) network. Among children born in 2006, in the 11 states included in the report, ASD prevalence at age eight was one in 59 children. In the early 1980s, only 35 years ago, the rate of autism was about one in 5000. That means that autism is about 85 times more common in today’s middle schoolers than it was in their parents. That is why nobody who grew up back then knew a child with autism but today everyone does.

The CDC report might as well be saying that one in 59 children has a hangnail for all the concern CDC expresses, when in fact, this is a public health disaster that will have repercussions for generations. At what point is the CDC going to admit that this disaster requires action? The largest amount by far of research dollars spent on determining causation has been directed towards genetics. “Genes don’t cause epidemics, environmental toxins do,” said World Mercury Project chair Robert F. Kennedy, Jr. “Why is the CDC doing nothing to identify the environmental toxins responsible for the most cataclysmic epidemic of our era?”

Even though vaccines are not mentioned, the above is a common antivaccine trope, particularly the part about how nobody who “grew up back then” knew a child with autism. In that, I like to point out that I took care of teens with severe autism when I worked at a group home one summer back in the early 1980s. I also like to point out that “environmental toxins” is antivaccine code for—you guessed it!—vaccines. To be fair, antivaxers don’t limit themselves only to vaccines as a potential cause of autism, but, make no mistake, any list of “environmental toxins” produced by antivaxers as causes for autism in essence consists of twenty entries of “vaccines” (or thimerosal), followed by everything else. Also in fairness, I actually do agree with the World Mercury Project that the US doesn’t spend enough on autism research. More importantly, the government doesn’t spend enough on autism services. None of that means that vaccines cause autism.

Elsewhere, Talk About Curing Autism (TACA), a group that promotes the idea that vaccines cause autism, as well as advocates dubious “biomedical” treatments (i.e., quackery) to “cure” autism notes:

The increase in prevalence rate cannot be explained by better diagnosis alone. Some have suggested that autism is just being better diagnosed today versus years ago and that many cases of intellectual disability are now being coded as autism. This would also assume that the experts diagnosing autism before did not know what they were doing. This is NOT TRUE. Autism is the only disorder dramatically on the rise while mental retardation or intellectual disability, Down syndrome and cystic fibrosis remain relatively the same.

It’s a common straw man argument that misrepresents the actual arguments being made as to how the “true” prevalence of autism probably hasn’t changed much in decades but diagnoses of autism have increased dramatically to use terms like “better diagnosis.” What is really meant is increased screening. I’ll return to an example I use so often: ductal carcinoma in situ (DCIS), a form of breast cancer. Well, actually, whether it’s really cancer or not is debatable, but it can clearly often be a precursor to cancer, although the percentage of DCIS lesions that progress to cancer isn’t precisely known. Be that as it may, before 1975 DCIS was a very uncommon diagnosis. Now it is very common. Indeed, back in the early 1900s, DCIS was rare because by the time it grew large enough to be a palpable mass, it almost always had become invasive cancer. Now, nearly forty years after mass mammographic screening programs became prevalent, DCIS is a common diagnosis. Indeed, approximately 40% of breast cancer diagnoses are DCIS, and a recent study found that DCIS incidence rose from 1.87 per 100,000 in the mid-1970s to 32.5 in 2004. That’s a more than 16-fold increase over a 30 year period, and it’s pretty much all due to the introduction of mammographic screening. This sort of thing should not be surprising to doctors, but apparently sometimes it is.

Unlike the case for autism, the diagnostic criteria for DCIS remained pretty stable over that 30 year period. Pathologists in the 1970s and pathologists in the 2000s would likely agree on what constitutes DCIS, which makes it particularly striking how, if you look for a disease or medical condition more intensely, you will always find more of it—often a lot more. Always. If this principle works for something that is diagnosed by an objective test, namely a biopsy, how much more so is it likely to be for a condition that has no unequivocal biochemical or tissue test to nail down the diagnosis, like autism, particularly for a condition whose diagnostic criteria has changed considerably over the last 25 years to be more inclusive? This brings up diagnostic substitution, in which diseases or conditions are classified differently now than they were 30 years ago, such that, for example, some children who might have been classified as mentally retarded 30 years ago receive an ASD diagnosis now. There is actually a fair amount of evidence for such a phenomenon, in which the decrease in diagnoses of intellectual disability are nearly completely offset by the increase in diagnoses of ASDs. There’s also the issue of whether we’re discussing administrative prevalence (diagnosis from school records of children receiving services) versus medical diagnoses of autism. (Hint: The two aren’t the same, and administrative prevalence is often higher, because the diagnosis of ASD is a basis for school-based services.)

None of this means that doctors 30 years ago didn’t know what they were doing. Diagnostic criteria were different, and there were nowhere near the resources in schools to identify potential cases.

A 13-year old “has more of a clue” than the CDC

None of this stopped the antivaccine propaganda blog Age of Autism from publishing a post by Cathy Jameson in which she uses her thirteen year old boy as an example of someone who “sees what experts can’t”:

Willem matter-of-factly added, “Well, we know what causes autism.”

Glancing back at his brother, I said, “Yep. We do,” and then added, “And you know what, Willem?”

“What, mom?” he asked.

“You know a heck of a lot more than some adults do,” I said proudly.

He said, “That’s good…but that’s kinda sad too, you know.”

“I know,” I told him. “It’s actually a little pathetic. Not about you knowing more – that’s going to help as you and brother get older. But the other people? It’s ridiculous how they’ve mislead the public about the rise in autism and what we know to be a cause of it.”

Looking back at his sisters, he said, “Yeah, they’d rather we celebrate it than make it stop.”

“Yep. That’s called turning a blind eye,” I said, “and I have a feeling that’ll continue to happen.”

He agreed.

Willem’s an insider looking out at what the world refuses to see—that vaccines can result in autism and that autism is here to stay. Since he knows that and since he sees the real-live struggle his brother lives through daily because of autism, he has every right to feel as frustrated as I do.

My 13 year-old has more of a clue than our paid government ‘experts’ do. If they ever fully understood just how much of a crisis autism is, I’d hope that the public would begin to pitch in a little bit more than know to do now.

Meanwhile, a website that is so extreme as to border on fake news, Big League Politics, is claiming that there was some sort of controversy and cover-up in the CDC that delayed the release of the report. The site claims that, rather than using the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSM-IV-TR), the 2018 report was going to use DSM-5, the newest version, which changes the diagnostic criteria for ASD. Of course, a clear reading of the text of the report shows that this is not the case and that the DSM-IV-TR served as the basis for case ascertainment. It is, however, true that the CDC does plan to phase out the use of the DSM-IV-TR with the 2016 numbers, to be published in 2020, but that’s only appropriate, given that the DSM-5 was released in 2013. I can see an argument for using the DSM-IV-TR for 2014 for the last time, but beyond that I have a hard time justifying it. Of course, antivaxers will likely argue that the change is being made to mask increases in autism prevalence, particularly if autism prevalence doesn’t continue its upward trend when the 2016 prevalence is published in 2020.

The autism epidemic that isn’t

One of the central narratives of the antivaccine movement is that we are in the midst of an “autism epidemic,” which I’ve sometimes heard called an “autism tsunami.” This narrative is essential to the idea that vaccines cause autism because of the expansion of the vaccine schedule in the 1990s that antivaxers blame for the “epidemic” that’s occurred over the last 25 years. Basically, no autism epidemic, no vaccine causation. Antivaxers know this. Unfortunately, what they either don’t know or refuse to acknowledge is that an increase in diagnoses of a condition is not necessarily the same thing as a true increase in prevalence. An increase in diagnoses without a true increase in prevalence (on a biological basis) can occur through, for example, changing of the diagnostic criteria over time or increased screening. My favorite example of the latter is, of course, DCIS, but there are others. For example, increased screening with ultrasound has led to a massive increase in the number of diagnoses of thyroid cancer, the vast majority of which are almost certainly overdiagnosis. Actually, there are issues with changing criteria for the diagnosis of thyroid cancer too. Indeed, diseases for which the prevalence is increasing often involve both increased screening and changes in the diagnostic criteria.

Another example is hypertension. Before the 1920s doctors didn’t routinely measure systolic/diastolic blood pressure ratios; so there were few, if any, cases of hypertension because doctors weren’t looking for it and didn’t know how to easily measure it. Even over the last decade, prevalence of hypertension has increased (for instance, in Canada). Perhaps a better marker for hypertension diagnoses is the percentage of adults who have been prescribed antihypertensive medications, which has skyrocketed since 1950. Of course, over that time, the definition of what constituted “hypertension” has changed markedly, to lower and lower diastolic and systolic pressures, as evidence showed that treating lower blood pressures could prevent premature death, although that trend appears to have leveled off. Amusingly, various denialists lambaste such broadening of diagnostic criteria to claim that hypertension is an “engineered” epidemic. And don’t even get me started on hypercholesterolemia.

So is the increase in autism prevalence “real” (as in an increase in the frequency of the neurobiological entity known as autism in the population) or is it an artifact of increased screening for school services, widening diagnostic criteria, and diagnostic substitution? There’s actually a fair amount of evidence that “true” autism prevalence has been fairly steady for decades. In other words, definitions of medical conditions matter. They can have a huge influence on prevalence rates observed. Moreover, if you don’t look for something, you won’t find it, and if you do look for something, you will find a lot more of it than you expected. It is quite possible that that is what’s happening with autism, but such a narrative doesn’t fit into the vaccine-autism idea. The antivaccine movement needs an increasing apparent-autism prevalence. If there’s no “autism epidemic,” there’s no way that vaccines cause autism. Over the years, there has been quite a bit of evidence that the “true” prevalence of autism has been fairly stable for quite some time. For instance, one study examined autism prevalence using stable diagnostic criteria between 1990 and 2010 and found “no clear evidence of a change in prevalence for autistic disorder or other ASDs between 1990 and 2010. Worldwide, there was little regional variation in the prevalence of ASDs.” Steve Novella has summarized several of these studies.

About a year ago, I took note of a new study out of Australia looking at the same criteria. It basically relies on one of my favorite sayings, namely that the harder you look for a medical condition the more of it you will find. Actually, it uses a corollary of that maxim, namely that, when you look for a condition and diagnose more of it (i.e., cause its apparent prevalence to increase), many of those cases will be milder cases. So, just as mammographic screening led to an increase in diagnosis of smaller tumors and premalignant DCIS, screening for prostate cancer led to milder cases of prostate cancer being diagnosed, and screening for hypertension led to milder cases of hypertension being diagnosed, if the increase in prevalence in autism is due to the broadening of the diagnostic criteria plus increased screening, we’d expect that the average severity of cases diagnosed would decline with time.

None of this rules out an increase in the “true” prevalence of autism over the last few decades, but the increasing consensus is that the evidence does rule out a massive increase. What is most likely happening is that we’re getting better and better at finding and diagnosing autism, including milder cases, to the point where the apparent prevalence of autism (i.e, the number of diagnoses) is starting to converge on the “true” biological prevalence of autism as it is defined today.