What’s in a number? 

Why does it seem so hard to figure out how many people have autism spectrum disorders (ASDs)? Those of us who have followed the various studies over the past 20 years are amazed and sometimes quite confused by the rising prevalence rates. Is there a worsening “epidemic” of autism cases? Are our children getting sicker and sicker? Two studies recently published may provide some clues to this puzzling problem. Before reviewing these new studies, a few words are in order about the dilemmas inherent in the strategies used for counting people with autism.

As most of us are familiar with ASDs, we acknowledge that there is a very wide range of capabilities, traits and symptoms that people with autism display. So how can one count people with this diversity of features? One needs to either virtually ask every individual in a defined population if they are on the spectrum, or if we assume that some individuals may not be aware they have the diagnosis, then examining every person in that population is the only way to be absolutely sure every person is counted.

Clearly this latter strategy is impractical. Instead, researchers have used various other methods to identify people likely to be on the spectrum. For example one study, done by the Centers for Disease Control (CDC; 2006), had trained teams review school and medical records to look for mention of ASD diagnoses or a collection of likely symptoms that might represent ASD. No children were actually examined. Findings suggested that both school and medical records under-counted children, but when both data sets plus measures of symptoms were combined, a rate of about 1 in 110 (0.9%) was calculated. Previous similar sampling techniques showed a lower prevalence rate of 1 in 150 (0.67%) (CDC; 2000).

This increase in apparent prevalence clearly begs the question, “Is the number of children with autism truly increasing?” Fortunately there are other alternative explanations to consider. For example, autism is increasingly covered in the media, younger children can now be diagnosed due to better techniques, and more teachers, parents and health care providers are screening for it. So one leading hypothesis is that there aren’t really more affected children, we’re just identifying them more readily for a variety of reasons.

Two more recent studies have been published, showing compelling results. Like prior CDC studies, these studies examined large entire populations; however, they went a step further and evaluated individuals in person. Thus these studies differ from the CDC studies, which did not actually examine and test individuals.

First let’s review a British study.1 Adults selected from a random sample in the U.K. were screened for autism risk using a questionnaire. Those with high risk were evaluated in person using the acknowledged single best test for ASD, the ADOS (Autism Diagnostic Observation Schedule). It was a massive study that tested, in person, over 600 individuals. In that group they identified only 19 people who met criteria for an ASD. A small number admittedly, but when all the math and statistical adjustments were made, low and behold, the rate of spectrum disorders was 1% or 1 in every 100 people, the virtually same rate as in children. Wow!

Just this week at the annual International Meeting for Autism Research (IMFAR) in San Diego, more details of this study were presented, which did highlight some methodological concerns. For example, the tools developed for diagnosis were designed for children; therefore, using them with adults requires further study. Also this study did not include severely affected or non-verbal adults, thus the actual number of adult individuals with ASD may have been higher. The authors acknowledged that this is still a preliminary estimate, but emphasized that to date, very little research has been done on this important population. Nevertheless, if this study is accurate, then there has not been an increase in ASD over time. Rather, ASD has been around at the same prevalence for a long time, only we were not adequately identifying it.

Now for the next surprise- a study was recently done in Korea using a similar strategy.2 First a “high risk” group was identified, that is, children in special education. Using current diagnostic criteria, the population prevalence in that group was 1 in 133 (0.75%), a bit lower than US studies. However, they went a step further and looked for possible autism cases in general education school programs. These were children without identified learning, behavioral or other concerning issues. Here were the surprising findings. They identified an additional 1 in 53 (1.89%) children aged 7-12 who met criteria for the broad autism spectrum. When you add this population together you get a prevalence of 1 in 38 or 2.64%. Again wow!

So how can we make sense of these rising numbers? It is hard. They indeed highlight the dilemma of identifying and counting a highly diverse population using different standards and techniques over time. Perhaps, indeed, there are many more children and adults on the spectrum than we had previously estimated. Perhaps we are counting using wrong criteria. Perhaps rather than an increasing “epidemic”, autism has always been part of the variation we all have as humans, but we are only now recognizing it.  Nevertheless, these numbers do demonstrate that we have a long way to go in providing the necessary services and supports to families, children and adults that will maximize their success and independence.

Stay tuned for additional blog posts about more fascinating research that was presented at IMFAR 2011.

1. Brugha TS, et. al. Epidemiology of Autism Spectrum Disorders in Adults in the Community in England,  Arch. Gen. Psychiatry. 2011; 68 (5): 459-66

2. Kim YS, et. al. Prevalence of Autism Spectrum Disorders in a Total Population Sample, Am J Psychiatry. 2011 (in advance of publication)