Q&A Conversations with experts about noteworthy topics in autism.
Julia Yellow
Opinion / Q&A

Questions for Stephen Blumberg: Tracking autism’s transience

by  /  3 November 2015
The Expert:

Stephen Blumberg

Associate director of science, National Center for Health Statistics

Estimating autism’s prevalence is tricky. No strategy is perfect, so researchers rely on a combination of approaches, such as sweeping national surveys and smaller pockets of meticulous monitoring.

These different tactics have all led to the same general conclusion: Autism is on the rise. Between 2012 and 2014, the prevalence rose from 1 in 88 to 1 in 68 children at age 8 — an increase of about 30 percent. But a new study from the U.S. Centers for Disease Control and Prevention (CDC) reveals an interesting twist. Roughly 13 percent of children with autism eventually lose their diagnosis, either because they outgrow it or because they never had autism to begin with.

The findings, published 20 October in Autism, highlight a pitfall of the tracking techniques that epidemiologists use. Surveys and medical-record reviews provide a snapshot of autism’s frequency that may fail to capture these ‘lost’ diagnoses, inflating the prevalence. There is also value in understanding which children lose their diagnosis and why.

We spoke to lead researcher Stephen Blumberg, associate director for science at the CDC’s National Center for Health Statistics, about the study’s findings and what they mean.

Spectrum: How does the CDC track autism prevalence?

Stephen Blumberg: The CDC monitors the prevalence of autism in two primary ways. Through the Autism and Developmental Disabilities Monitoring Network, experts review medical and educational records from select sites across the U.S. to identify children who have an autism diagnosis or behaviors that would strongly suggest an autism diagnosis. It is in many ways considered a gold-standard measure of autism prevalence, but it takes a long time to compile that information. The most recent report — the one that says 1 in 68 children have autism spectrum disorder — is based on 2010 data for 8-year-old children in 11 communities.

The other way is through nationwide surveys that ask parents whether their children have ever been diagnosed with autism spectrum disorder by a doctor or other healthcare provider. The surveys have large sample sizes and are of high quality, but are limited by the knowledge parents have and their ability to recall their children’s diagnoses. In addition, some children with autism may not have received a diagnosis from a healthcare professional because they have limited access to the healthcare system, or they’re in a school system that has limited resources to conduct or share diagnostic evaluations.

Each system has its strengths and weaknesses. Together, they help us to understand the prevalence of autism, and both show that the prevalence has been increasing over time.

S: Why did you conduct this study?

SB: I think researchers have recognized that autism diagnoses sometimes change due to misdiagnoses, maturation or treatment. The purpose of our study was to better understand which of those reasons is most common and to learn who’s most likely to lose an autism diagnosis.

S: What did you find?

SB: Approximately one out of eight children ever diagnosed with autism had lost their diagnosis. And of the children who had been diagnosed with autism but no longer had a diagnosis, the parents of 74 percent said a doctor changed their child’s diagnosis because of ‘new information.’ This translates to 9 percent of school-age children ever diagnosed with autism having lost their diagnosis due to new information.

S: What is this new information?

SB: We don’t have a clear answer. We didn’t ask what the new information was or what guided the doctor’s decision-making. What we did find is that when a doctor changed the diagnosis, attention deficit hyperactivity disorder (ADHD) was the most common replacement diagnosis. But children who lost their autism diagnosis were no more likely to have ADHD than children currently diagnosed with autism.

S: Why ADHD?

SB: Symptoms of hyperactivity, impulsivity and inattention are common in children who have autism. For some children with ADHD, these symptoms may initially receive less attention because of social communication issues that they eventually grow out of.

S: What were some other reasons for a lost diagnosis?

SB: Sometimes a doctor may give an autism diagnosis so that a child with other developmental issues can receive necessary services. In about one in four children who lost an autism diagnosis, the parents said the initial diagnosis was given for this reason. In addition, for about one in five children, the parents gave an answer suggesting that early intervention, treatment or maturity resulted in the child’s behaviors changing or the autism symptoms going away.

S: What do your findings say about the rise in autism?

SB: Our finding that autism diagnoses are not necessarily permanent can complicate the use of surveys to estimate the prevalence of autism. However, our estimated prevalence of lost diagnoses — about 13 percent of children who are diagnosed — is still too low to explain the recent increase in the estimated prevalence of autism.

S: Should researchers change the way they track autism prevalence?

SB: As a result of the findings here, we’re adding a question about current autism [diagnosis] to the National Health Interview Survey — a representative survey of more than 40,000 households in the U.S. This annual survey now asks only whether a child has ever been diagnosed. By adding the new question, we can use that survey to monitor the difference between children who have ever been diagnosed with autism and those who currently have a diagnosis.

  • Frank Kelly

    Until we have reliable biomarkers, the DSM definitions of “Autism” and “ADHD” are just a measure of visible symptoms.

    Sadly the result of this news is that some people will now believe ASD is just “over diagnosed” when in fact the result could be because diagnosing ASD is really hard and/or some kids achieve an “optimal outcome” and their symptoms become subclinical.

    My experience in this community tells me for every 1 ASD kid with a changed outcome – there are 2-3 kids who have the wrong diagnosis (e.g. ADHD or ODD) or NO diagnosis and so the numbers we have are an underestimate.

    • Hi Frank,

      Thanks for your comment. I asked Dr. Blumberg about the relative frequency of underdiagnosis and overdiagnosis. Here’s what he said:

      Our study using parent-reported survey responses does not address the question of underdiagnosis, that is, the percentage of children with ASD who do not have an ASD diagnosis. Therefore, we cannot say whether underdiagnosis occurs more or less often than overdiagnosis. However, the most recent report from the Autism and Developmental Disabilities Monitoring Network finds that only 80% of all children meeting their ASD surveillance case definition had either eligibility for autism special education services in their education records or a DSM-IV or ICD-9 diagnosis documented in their medical records.

  • EJ

    Autism is a lifelong condition, you cannot outgrow it (http://www.forbes.com/sites/emilywillingham/2013/01/17/can-people-really-grow-out-of-autism/) and losing your diagnosis (I note the careful wording you used) is not the same as no longer having autism. People with autism may learn to mask their difficulties or overcome some with learning (but it will not have been instinctive as in a neurotypical) but this is not without great ongoing effort and resultant burnout. Look at your own article on the supposedly “mildest” form of autism and the effects on mental health https://spectrumnews.org/news/people-with-milder-forms-of-autism-struggle-as-adults/ I agree there will be some misdiagnoses in the first place, with children being diagnosed ever earlier, there is the chance for some other developmental issues or delays being misconstrued. Also there is this: http://www.nbcnews.com/health/kids-health/study-suggests-autism-being-overdiagnosed-n450671 and some doctors have admitted to misdiagnosing autism to assist parents get access to supports for their child. This is very wrong, but in America where people pay for a lot of their assessments and treatments there is a customer mentality so people are more likely to be given what they are paying for. In the UK autism is likely to be underdiagnosed as there is still not sufficient expertise on high-functioning autistic conditions and autistic females as a group are vastly underdiagnosed anyway. Some parents are told by CAMHS that unless their child is causing disruption at school there is no value in a diagnosis and to avoid a drain on services they withhold diagnoses, using the “not enough traits for a diagnosis” lark. You can’t be a little bit diabetic, or pregnant, or dead any more than you can be a little bit autistic. And why do I still keep seeing figures as being 1 in 68 when it’s supposedly even higher at 1 in 50 in the USA? http://www.nvic.org/NVIC-Vaccine-News/April-2013/CDC-Reports-1-in-50-American-Children-Diagnosed-wi.aspx As it’s cited to be about 1 in 100 in the UK, clearly there is a discrepancy, with some overdiagnosis in the US and some underdiagnosis in the UK. Either that or there are significantly more environmental factors exacerbating this side of the epigenetic formula in the USA. Autism is on the increase, however statistics are manipulated and you should look into epigenetic inheritance also as an explanation as to how vaccinations impact the mother’s immune system and those effects are carried forward onto her child. This would account for at least some of the cases where autism appears in non-vaccinated children. Look also at the discrepancy in rates between Amish and non-Amish populations.

    • Mark Carew

      There is a need to identify the issues complicating public medicine programs , including vaccination. But it is a thankless task for the professional who will always be ridiculed for voicing a dissenting perspective or even hinting at it. This sad fact stifles debate and confines competent understanding of the issues to private quarters. There is no universal truth, what is true for one is not true for every other. Mass vaccination programs and all public health expenditure is simply a product of the State attempting to maximize efficiency of its population. Without it, the State would likely loose ground relative to neighbors, we see this regularly in the modern influenza programs hoping to save days lost. Unfortunately some individuals, with their unique DNA heritage will be susceptible to harm from the vaccine, or the vaccine batch may have some unexpected legacy issue. Yet the State is obliged to mute any debate surrounding the efficacy of the program for fear of panic reaction, and a breakdown of the immunity level in the population and so routine epidemics. Autism in some cases is inevitably linked to a vaccination event. Sensible scientific understanding of this fact leads one to understand something about vaccination, virus, and our unique DNA. The gender bias we see is common to many heritable conditions, including say Insulin Dependent Diabetes, among may others. This illustrates that one purpose of the individual male creature has to do with his environmental capabilities. The male reaching reproductive maturity has a chance to pass on his genetic heritage, only if the specific environment concurs.(this is an important factor in understanding gene pool drift and rising autism and diabetes type 1) Conversely the female counterpart preserves recessive genes from this trial, and all her offspring have a chance to inherit the recessive , good or bad. One effect of mass vaccination which may not be intended is the survivorship of less than competent genes in male offspring. Becoming viable they are able to replicate and complement the gene pool generally. Of course this knowledge has no value for a modern liberal democracy, where parents are obliged to consent vaccination or trust to natural if premature disease. Much better to embrace the truth than torment parents with denial and untruth.

  • Ethyl

    When asked if their child had ~ever~ had an autism diagnosis, the numbers went from 1 in 68 to 1 in 45 children in the 2015 CDC survey.That’s an increase of 150% in ONE YEAR that they attributed to the way they asked the question, “Has your child EVER been labelled as ASD.” This, to me means, that 23 of the 68 had an autism label and lost it.

    Way, way, way beyond what is attributed here…23 of 65 means fully one third lost the diagnosis.


Log in to your Spectrum Wiki account

Email Address:



Request your Spectrum Wiki account

Spectrum Wiki is a community of researchers affiliated with an academic or research institutions. To be considered for participation, please fill out this form and a member of our team will respond to your request.


Email Address:

Title and Lab:

Area of Expertise: