This is a follow up to a previous post, Understanding CDC Prevalence Estimates.
It is a popular misconception that there is a growing autism epidemic in the US based in part on the changing prevalence numbers reported by the CDC. But given that the CDC is monitoring a sample of those most likely to develop autism one cannot make the claim that there is an epidemic. To the contrary, it is still unclear how many autistic people there are. The most accurate conclusion that can be drawn about the size of the population of autistic individuals using the changing CDC prevalence data is that more of the children most likely to have a developmental disorder at the ADDM sites are being diagnosed and receiving treatment.
So what can the CDC prevalence numbers tell us if they can’t be generalized to the whole population? Plenty. First, we see that race and ethnicity influences identifying ASD and also IQ placement. Second, while it is established in the scientific literature that ASD can be diagnosed as early as 12 months of age it is more commonly identified at 53 months. This suggests that something is causing a delay in diagnosis. Third, boys are still more often diagnosed than girls. Fourth, despite developmental concerns being noted for a number of the children seen at the monitoring sites these concerns did not always generate a comprehensive evaluation for the child in question.
What can parents take away from this discussion of CDC prevalence numbers? Take a second critical look at those making claims about there being an epidemic of autism because, frankly, the numbers available to us do not support such claims. If you as a parent have concerns about your child’s developmental progress push for a comprehensive evaluation of your child. Don’t be put off or take a wait and see approach.
In the United States the Centers for Disease Control and Prevention (CDC) maintains a network of 11 autism monitoring sites across the country known as the Autism and Developmental Disabilities Monitoring (ADDM) Network. Published numbers on prevalence from the CDC come from observations at these sites. The CDC typically publishes the latest prevalence numbers available in the Morbidity and Mortality Weekly Report (MMWR) Surveillance Summaries. This publication is available for viewing and download on the CDC website www.cdc.gov. The prevalence estimate published in March 2014 (1 in 68, or 14.7 per 1,000) was based on ADDM network observations from 2010.
First let’s define some terms. You may hear in the terms prevalence and incidence used interchangeably in the numbers discussion surrounding autism. However, prevalence and incidence are not the same. There are two types of prevalence to be aware of in this discussion. Point prevalence is a ratio representing the portion of a population that has a condition at a single point in time, point x. Period prevalence is a measure of the portion of a population that has a condition in a set span of time, period x to y. Incidence is the rate at which new cases of a condition occur per person over a period time. Cumulative incidence is the proportion of a population that became new cases within a specified time period. Research is generally focused on point and period prevalence of autism spectrum disorders.
The CDC has been monitoring ASD through the ADDM network since 2000 The following are the criteria, as reported in the MMWR Surveillance Summaries (pp. 3, 4), for inclusion in the cohort samples monitored by the CDC at the ADDM sites.
1. Birth year (children 8 years old during surveillance period) and residency. (children must live in the areas covered by the ADDM sites during the period of surveillance). This is the age by which diagnosis of a developmental disorder is most likely to have been made.
2. Eligible for special education services, have an International Classification of Diseases, Ninth Revision (ICD 9) disability or psychological condition.
3. The presence of certain behavioral or diagnostic descriptions in the children’s files.
4. A diagnosis of disability upon review by trained clinicians blinded to any identifying information about the children based on DSM-IV-TR criteria.
The first criteria, birth year and residency, define the population from which the CDC draws its samples. From this population a sample is taken, based on the remaining criteria, to determine the prevalence of autism spectrum disorders. This selection process presents one of the first problems with the CDC prevalence data. The sample of children monitored by the CDC is not representative of the population of children in the US. The children monitored at the ADDM sites are chosen because they have a greater likelihood of having a developmental disorder. Furthermore, the ADDM sites themselves are not randomly distributed and do not provide a true representation of the US population. This biased sample means that the prevalence numbers reported by the CDC cannot be used to make inferences about the population as a whole. While the CDC makes it clear that these are limitations to their surveillance data these points seem to go ignored by the public.
See the follow up post, CDC Prevalence Estimates and ASD Epidemic.