Autism Prevalence: More Affected or More Detected?

Tom Insel, director of the U.S. National Institute of Mental Health and Chair of the Interagency Autism Coordinating Committee has published a blog article coinciding with the CDC announcement of 1 in 88 estimated prevalence.

As a government publication, I feel it is OK to copy it here in total, but you are encouraged to read it on the NIMH website: Autism Prevalence: More Affected or More Detected?

Autism is always surprising. Earlier today, the CDC released new numbers from their ongoing surveillance of autism prevalence, the Autism and Developmental Disability Monitoring (ADDM) Network. What was once considered a rare disorder is now reported as affecting 1 in 88 children, 1 in 54 boys. These new numbers, up 78 percent from 2002 and 23 percent from 2006, raise immediate questions. Are more children affected or more detected? Does the increase reflect a growing problem, or do these new numbers reflect an improvement in our ability to diagnose and serve those affected?

These new data do not answer these questions. The CDC surveillance project focuses on 8-year-olds identified in 2008; that is, children who were born in 2000. By definition, autism begins before age 3, so a focus on 8-year-olds should capture anyone who was identified and still has a diagnosis. The prevalence numbers are based on a standardized assessment of descriptions of behaviors culled from administrative or health records from select communities, not on standardized diagnostic interviews in the general population. The strength of this approach is its wide reach, allowing a comparison across 14 states. The CDC reports a four-fold variation across sites. These new results, as with those from other records-based surveillance systems, do not answer questions related to why the identified prevalence of autism has changed over time.

Other research suggests a complicated picture. A total population study of all 7–12-year-olds in a town in South Korea (more than 55,000 children) used standardized diagnostic instruments for children who screened positive and reported a prevalence of 2.64 percent.1 That is 1 in 38 children! There is no reason to expect that this prevalence is unique to this community. To be sure, two-thirds of these children had never received a diagnosis of autism spectrum disorder (ASD), meaning that the identified prevalence was closer to 1 percent or one in 100, roughly the same prevalence reported in the United States. From this perspective, the increase reported by the CDC might mean we are better at detecting children who meet criteria for ASD, but potentially we still are only halfway to the actual prevalence in the general population. Indeed, the biggest increase in the CDC surveillance report was in Hispanic and African American children, groups which previously had low rates of detection.

But can we be certain that more children are not affected? Data from the Developmental Disability Services registry in California demonstrates a 12-fold increase in the number of children receiving services for autism over the past 20 years, with a continuing rise recently. But these data, while dramatic, cannot rule out increased use of the diagnosis. Bearman and colleagues who have studied the California trends suggest that only about 26 percent of this increase can be explained by diagnostic substitution, especially for the most severe cases — children with intellectual deficits — which may not have been identified as autism in an earlier era. Another fraction can be attributed to better ascertainment or detection especially for children at the less severe end of the spectrum. Together these factors explain only a part of the linear increase observed in the California registry. In the absence of other explanations they and others suggest that a real increase is quite likely.2, 3, 4

Which makes a recent report from England especially surprising. In a careful epidemiological study based on a national sample (n = 7,461 adults) from 2007, Brugha and colleagues did careful diagnostic assessments based on standardized interviews. They found that familiar rate of about 1 percent in adults across the entire age range without a significant reduction in the older part of the sample as one would expect if the prevalence had increased in recent years.5

This takes us back to the central question: has the number of children with ASD increased or not? Total population epidemiological studies suggest much or all of the increase is due to better and wider detection. Studies of administrative and services data suggest that better detection cannot fully explain the profound and continuing increase. Are we seeing more affected or more detected? The question is vitally important, but there is not one, simple answer just as autism is not a single, simple disorder.

If there is an increase in the number affected, then we need to find the causal factors to bend the curve. Analogous increases in food allergies, asthma, and Type 1 diabetes have provoked an aggressive search for environmental causes. If the number of children with ASD has not changed, but we are diagnosing and serving 12-fold more of them over the past two decades, then we need to focus on better diagnosis and treatments rather than looking for new environmental factors driving the precipitous increase.

Science can resolve this dilemma, but the methods to examine this question as well as the answers will be complex. While it is never possible to go back in time, longitudinal population based studies and even careful retrospective studies can determine if more children are affected and if the nature of the disorder is changing over time. The changes in prevalence of other developmental disorders, measured with biomarkers (Type 1 diabetes) or emergency room visits (food allergies), appear to be true increases in the number of children affected. As diagnostic changes and ascertainment fail to explain the majority of the increase in autism prevalence, it seems prudent to assume that there are indeed more children affected and continue an aggressive search for causes while striving to improve detection, treatments, and services. Our working assumption is that there are both more children affected and more detected.

References

1Kim YS, Leventhal BL, Koh YJ, Fombonne E, Laska E, Lim EC, Cheon KA, Kim SJ, Kim YK, Lee H, Song DH, Grinker RR. Prevalence of autism spectrum disorders in a total population sample. Am J Psychiatry. 2011 Sep;168(9):904-12. Epub 2011 May 9. PubMed PMID: 21558103.
2King M, Bearman P. Diagnostic change and the increased prevalence of autism. Int J Epidemiol. 2009 Oct;38(5):1224-34. Epub 2009 Sep 7. PubMed PMID: 19737791; PubMed Central PMCID: PMC2800781.
3Keyes KM, Susser E, Cheslack-Postava K, Fountain C, Liu K, Bearman PS. Cohort effects explain the increase in autism diagnosis among children born from 1992 to 2003 in California. Int J Epidemiol. 2011 Dec 7. [Epub ahead of print] PubMed PMID: 22253308.
4Bresnahan M, Li G, Susser E. Hidden in plain sight. Int J Epidemiol. 2009 Oct;38(5):1172-4. PubMed PMID: 19797336.
5Brugha TS, McManus S, Bankart J, Scott F, Purdon S, Smith J, Bebbington P, Jenkins R, Meltzer H. Epidemiology of autism spectrum disorders in adults in the community in England. Arch Gen Psychiatry. 2011 May;68(5):459-65. PubMed PMID: 21536975.