What's Behind the Rise in Autism?
An excerpt from FRONTLINE's interview with Dr. Eric Fombonne
- Regressive autism
- Is there an epidemic?
- California's autism numbers/the issue of diagnostic substitution
He is a child and adolescent psychiatrist and an epidemiologist at McGill University and has specialized in researching the epidemiology of mental illness, with a focus on the factors influencing the increase in the diagnosis of autism. This is the edited transcript of an interview conducted on Dec. 21, 2009.
Classically, how did you diagnose autism three or four decades ago?
Three or four decades ago, you needed to have a child who didn't develop well in terms of his communication skills. That involved usually language, which was delayed, and when it was developing, it was developing with abnormal patterns of use. [These children] tended to create new words, had difficulty to understand, who are repeating words on and on without meaning or without communicative intent.
They also had failure to develop relationships in a reciprocal sense, so they didn't play well with other kids, they tended to be solitary, in their own world. And with play, they tended to develop unusual patterns of plays, largely [marked] by a tendency to look at features of the environment or toys, not look at the toy as a whole. They failed to develop imaginative play, which is this critical aspect of child development by which a child engages into imagination, develops scenario and script which are viable and lively and imaginative.
This was a pretty rare condition back in those days?
Yes, in those days. In the 1960s, the view that autism was a very deviant condition, very rare, very abnormal; and the prevalence of autism at that time was estimated to be about 4 or 5 children for 10,000 children. So something which would be considered to be very rare.
Regressive autism is not a particular diagnosis. It's a form of onset of autism, in which parents usually see the child developing some words and then the child seems to lose the words he was using in the previous month or the previous week. So there is a loss of skills, usually occurring in the end of the second year of life. And that has been actually known as early as Kanner.
Editors Note: Autism was first described by Dr. Leo Kanner in 1943, who wrote about 11 children who exhibited an apparent congenital lack of interest in other people.
If one reads the descriptions of Kanner in 1943, there were actually descriptions of children who lost their skills, lost their language in the second year of life.
But at that time there was no particular attention drawn on this subset of children with autism, which applies to about 20 or 30 percent of them.
A common argument says there's an epidemic of autism and that vaccinations may be responsible for the epidemic. How long have you been hearing these arguments?
I have heard these arguments for about 12 years. And I think it really started in the late 1990s, out of this hypothesis that the measles vaccines would be responsible for an increase in the numbers of children diagnosed with autism. So it was all linked together.
Truly, if one looks at the epidemiology of autism or the number of children diagnosed with autism, we know that in the 1990s, in most parts of the world, there was a clear trend for an increasing numbers of children to be diagnosed with a condition called now pervasive developmental disorder or autistic spectrum disorder [ASD].
That trend was seen in most countries as early as the 1990s. And it's only in the years later that some people tried to explain that trend and link that increase in numbers of children diagnosed with vaccine practices.
It depends what we mean by that. If we look at the prevalence, as published in various journals, yes, the prevalence has increased.
We were looking at a prevalence in the early 1960s which was 4 or 5 children affected amongst 10,000 children. And now, the rate we commonly agree upon is close to 1 percent -- so 100 [children] per 10,000, which is about 20-25 [times] more than what we had 45 years ago. So what does that mean?
To claim that there would be an epidemic, it would need to have [a certainty] that we are comparing like with like. And in fact, when we are comparing these rates, we are not comparing the same definition of the disorder.
In the 1960s we had this very narrow definition of autism -- the definition that was used by Kanner with children who are very abnormal in their development, completely qualitatively deviant in many aspects of their functioning. They had deficits which were pronounced in social functioning, in language and communications and play. So they were the severe form of autism that we still see, but which is just one part of what we now call autism.
In fact, when we look at the recent prevalence studies and the recent rates that we quote, like 1 percent, we actually look at a spectrum of condition which was not at the time investigated by the those [earlier] studies.
So today when we go and set up studies to look for autism in a population, you have to have a definition that you're going to use. And the definition we use is one -- PDD or pervasive developmental disorder -- that is used in the DSM IV and which contains different forms of the disorder.
The severe form of autism that was in use in Kanner's time is now just one part of autism. So there are children now who have a diagnosis of autism who are actually normally intelligent, have no mental retardation associated with the condition, and have good language skills. And they were not accounted for in the earlier studies.
And alongside this more broadened spectrum of autistic disorder, we have also atypical forms of autism. These are children who are not meeting full criteria for autistic disorder, but have less severe symptoms or an onset which is a bit later. And we have also a newcomer in the diagnostic system -- Asperger's syndrome -- which was not included at all in the early studies.
So we are now talking, in effect, about a much broader spectrum of condition, which includes different subtypes of autism, and which includes phenotypes which are much milder than what we were starting with in the 1960s.
So there is no surprise therefore that the rate would be higher, because we have actually refined our concept. But we have actually broadened the concept, and children who were not diagnosed in this spectrum 40 years ago are now diagnosed on it. So of course the rate is higher.
Yes. This broadening of the condition means that some children would be diagnosed with these autism spectrum conditions now and not 40 years ago. So the question is: But what were they diagnosed with, or where were they at the time they were diagnosed with autism? We have now strong evidence that this broadening of this definition and the enlargement of our concepts of autism have actually led to this broadening of the prevalence rate.
So for instance we have several data showing that children who were previously diagnosed with mental retardation are now, or would be now diagnosed with an autism spectrum diagnosis. There are very good data published in the U.S. -- looking at the 50 states and the special educational data sets from the U.S. -- which show that there has been, over time, as autism numbers were increasing ... a decrease in the number of children diagnosed with mental retardation.
More recent studies which have looked at individual trajectories in California have documented that in fact if you follow children during that period of the 1990s, they were initially entering the system of public service with a diagnosis of mental retardation, and would, as time elapsed, would earn a diagnosis of autism at a time where the policy changed about what it meant to have an autism diagnosis, and it was actually an opening door for more services.
So this recent study in California indicates that probably 25 percent of the increase in this California state of public services might be due to this shift from an initial diagnosis of mental retardation to an autistic disorder diagnosis. So that's a significant proportion of the increase, which is explained by the substitution. And it's very legitimate to postulate that there has been other pathways for diagnostic substitution.
We actually know from British studies that some children who were in the past diagnosed with language disorders would now be -- when they are reviewed with more modern tools -- they would be falling into the spectrum and would be most certainly diagnosed with an autism spectrum condition.
A study in the UK showed that very well-characterized children and adolescents with language disorders, when they are reviewed 10 years later with tools and diagnostic assessments which are looking at autism in these children, actually would identify autism in 66 percent of them. Two-thirds of them would be now considered as meeting criteria for the condition. So that's another pathway from a previous diagnosis to this diagnosis of autism.
And by clinical experience, we can certainly say that some of the children who were previously diagnosed with obsessive compulsive disorder, schizotypal personality disorders, and other kinds of unusual psychiatric disorders might now be regarded as having autism.
So most certainly, this diagnostic substitution hypothesis has applied from a range of other conditions, and is one of the pathway through which now we have an increased number of children diagnosed with ASD.
So there are many, many reasons why the numbers have gone up, which do not necessarily call for an epidemic. …
It remains a possibility that, in addition to the factors I described -- the diagnostic broadening, the concepts which have changed and are being enlarged, and better identification of children in populations -- there are other factors which might also play a role and might indeed be changing the incidence of the condition and contributing to this increase. But if this is the case, we don't know what they are, and they certainly are not vaccines.