A new study sheds more light on the question of what is causing the recent increase in the rate of diagnosis of autism. Professor Dorothy Bishop from the University of Oxford studied adults who were diagnosed in 1980 with a developmental language disorder. She asked the question – if these people were subjected to current diagnostic criteria for autism, how many of them would be diagnosed today as having autism? She found that 25% of them would. (Bishop 2008)

This epidemiological question has been at the center of a controversy over whether or not there is a link between vaccines (or the mercury-based preservative, thimerosal, that was previously in routine childhood vaccines) and autism. The primary evidence for this claim put forward by proponents of a link is that the number of diagnoses of autism increased dramatically at the same time that the number of vaccines routinely given to children was increasing in the 1990’s. They are calling this rise in autism an “epidemic” and argue that such an increase requires an environmental factor, which they believe is linked to vaccines.

That the number of new autism diagnoses is dramatically increasing is generally accepted and not a point of debate. The historical rate of autism is about 4 per 10,000 and the more recent estimates are in the range of 15-20 per 10,000 (30-60 per 10,000 for all pervasive developmental disorders of which autism is one type). (Rutter 2005) The controversy is about what is causing this rise in diagnoses. There are two basic hypotheses: 1) That the true incidence of autism is rising due to an environmental cause, 2) That the rise in incidence is mostly or completely an artifact of increased surveillance and broadening of the definition of autism. These two hypotheses make specific predictions, and there is much evidence to bring to bear on their predictions – this recent study only being the latest.

The confusion about the epidemiology of autism is one common to scientific medicine. Whenever historical comparisons are made it is possible that changing definitions and practices over time will distort those comparisons. This is why medical scientists are often reluctant to change nomenclature (disease names) and definitions – doing so immediately renders the literature obsolete. All subsequent literature must now have a footnote. But the progress of our understanding of biology and disease makes such changes unavoidable.

In the 1990’s the diagnosis of autism was changed to autism spectrum disorder (ASD) – the new name reflecting the changing concept of autism to include a broader spectrum of symptoms, including much more subtle manifestations. In particular a diagnostic entity known as Aspergers syndrome, which is essentially a subtle manifestation of autism features, was classified as part of ASD. Any time you broaden a category the number of individuals that fit into that category is likely to increase.

Autism researcher Eric Fombonne found that:

Recent epidemiological surveys of autistic disorder and other PDDs have heightened awareness of and concern about the prevalence of these disorders; however, differences in survey methodology, particularly changes in case definition and case identification over time, have made comparisons between surveys difficult to perform and interpret. (Fombonne 2005)

In addition to the broadening of the diagnosis, the social and medical network supporting ASD dramatically increased. There has been increased efforts at surveillance – scouring the community for hidden cases of autism. Further, parents have become much more accepting of the diagnosis, which may partly be due to the fact that is some states the label with facilitate access to special services. And clinicians have become more knowledgeable of ASD so are better able to make the diagnosis, even in subtle cases.

Rutter, in order to test this latter hypothesis that increased diagnostic rates were due largely to changes in diagnosis and surveillance, reviewed literature that contained sufficient information to assess true historical rates of autism. He found that applying modern criteria to these historical records yields similar rates of diagnoses: 30-60 per 10,000. Taylor did a similar review and found the following:

The recorded prevalence of autism has increased considerably in recent years. This reflects greater recognition, with changes in diagnostic practice associated with more trained diagnosticians; broadening of diagnostic criteria to include a spectrum of disorder; a greater willingness by parents and educationalists to accept the label (in part because of entitlement to services); and better recording systems, among other factors. (Taylor 2006)

Another prediction that flows from the second hypothesis is that if we compare apples to apples – meaning if we look at the same community and apply the same diagnostic methods that were used in the past as documented in a published study, then the incidence should be the same. In other words – if we control for any changes in the diagnostic criteria and surveillance methods the incidence of autism should be stable over time. Chakrabarti and Fombonne did exactly that, comparing the incidence of autism in 2002 (looking at a cohort of children born between 1996-1998)  to the same population using the same methods as a previous study looking at the cohort of children born between 1992-1995. They found:

The rate in this study is comparable to that in previous birth cohorts from the same area and surveyed with the same methods, suggesting a stable incidence. (Chakrabarti 2005)

If the broadened diagnosis hypothesis is true than it must also be true that as other diagnoses shifted over to autism they would decrease as autism numbers increased. This is exactly what Jick et al found when they reviewed a cohort of boys with and without autism. What was previously diagnosed as language disorder is now being diagnosed as autism, with a corresponding decrease in non-specific language disorders. Shattuck found the exact same effect, so called “diagnostic substitution,” when he studied the prevalence of disabilities among children in US special education from 1984 to 2003. He found that in locations where the prevalence of autism had increased there was a corresponding decrease in the prevalence of other disabilities. (Shattuck 2006)

This brings us to the current study by Bishop et al. They looked at adults who were diagnosed as children with a developmental language disorder. Applying current diagnostic criteria they found that a quarter of them (12) would have been diagnosed today with ASD. This also fits the hypothesis that diagnoses have been shifting over to autism from other developmental disorders over the past two decades. The authors, however, caution that the number of subjects in their study was small and therefore it is difficult to extrapolate from this to the general population.

It should also be noted that all of this research, while supporting the hypothesis that the rise in autism diagnoses is not due to a true increase in the incidence but rather is due to a broadening of the definition and increased surveillance, does not rule out a small genuine increase in the true incidence. A small real increase can be hiding in the data. There is no evidence upon which we can conclude, however, that true autism rates are increasing.

Of course the implications of this are profound. If there is no autism epidemic, if there is a “stable incidence” of autism over recent decades, then this alone is powerful evidence against the vaccine hypothesis – and in fact removes the primary piece of evidence for a vaccine-autism connection. Just as a true increase in incidence would have called out for an environmental factor causing autism, the lack of any increase argues strongly against any environment factor – especially when this is combined with the copious evidence for multiple genetic factors as the ultimate cause(s) of ASD.



Bishop DV, Whitehouse AJ, Watt HJ, Line EA. Autism and diagnostic substitution: evidence from a study of adults with a history of developmental language disorder. Dev Med Child Neurol. 2008 Mar 31

Chakrabarti S, Fombonne E. Pervasive developmental disorders in preschool children: confirmation of high prevalence. Am J Psychiatry. 2005 Jun;162(6):1133-41.

Fombonne E. Epidemiology of autistic disorder and other pervasive developmental disorders. J Clin Psychiatry. 2005;66 Suppl 10:3-8.Click here to read

Jick H, Kaye JA. Epidemiology and possible causes of autism. Pharmacotherapy. 2003 Dec;23(12):1524-30.

Rutter M. Incidence of autism spectrum disorders: changes over time and their meaning. Acta Paediatr. 2005 Jan;94(1):2-15.Click here to read

Paul T. Shattuck. The Contribution of Diagnostic Substitution to the Growing Administrative Prevalence of Autism in US Special Education. PEDIATRICS Vol. 117 No. 4 April 2006, pp. 1028-1037

Taylor B. Vaccines and the changing epidemiology of autism. Child Care Health Dev. 2006 Sep;32(5):511-9.Click here to read


Posted by Steven Novella

Founder and currently Executive Editor of Science-Based Medicine Steven Novella, MD is an academic clinical neurologist at the Yale University School of Medicine. He is also the host and producer of the popular weekly science podcast, The Skeptics’ Guide to the Universe, and the author of the NeuroLogicaBlog, a daily blog that covers news and issues in neuroscience, but also general science, scientific skepticism, philosophy of science, critical thinking, and the intersection of science with the media and society. Dr. Novella also has produced two courses with The Great Courses, and published a book on critical thinking - also called The Skeptics Guide to the Universe.