Evaluation of a Records-Review Surveillance System Used to Determine the Prevalence of Autism Spectrum Disorders.

28 Jun

Perhaps the most well-known statistics about autism come from the CDC’s autism surveillance project. The Autism and Developmental Disabilities Monitoring (ADDM) Network is a multi-state operation coordinated by the CDC. Their reports are the source of the “1 in 166”, “1 in 150” and “1 in 100” numbers one hears in the press and elsewhere. The Netweork looks at children’s medical records, educational records and sometimes both sets of records and make determinations of which children have autism spectrum disorders (ASD’s). This allows the Network to monitor large numbers of children. But they don’t actually test each child,bringing up questions of how accurate are the prevalence numbers.

A CDC team has looked into this question (actually, questions) in a recent paper:

Evaluation of a Records-Review Surveillance System Used to Determine the Prevalence of Autism Spectrum Disorders.

Nonkin Avchen R, Wiggins LD, Devine O, Van Naarden Braun K, Rice C, Hobson NC, Schendel D, Yeargin-Allsopp M.

Developmental Disabilities Branch, Centers for Disease Control and Prevention, 1600 Clifton Road, MS E-92, Atlanta, GA, 30333, USA, ravchen@cdc.gov.

We conducted the first study that estimates the sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of a population-based autism spectrum disorders (ASD) surveillance system developed at the Centers for Disease Control and Prevention. The system employs a records-review methodology that yields ASD classification (case versus non-ASD case) and was compared with classification based on clinical examination. The study enrolled 177 children. Estimated specificity (0.96, [CI(.95) = 0.94, 0.99]), PPV (0.79 [CI(.95) = 0.66, 0.93]), and NPV (0.91 [CI(.95) = 0.87, 0.96]) were high. Sensitivity was lower (0.60 [CI(.95) = 0.45, 0.75]). Given diagnostic heterogeneity, and the broad array of ASD in the population, identifying children with ASD is challenging. Records-based surveillance yields a population-based estimate of ASD that is likely conservative

If you are looking for a glossary for these terms, one can be found here.

Sensitivity: Proportion of persons with condition who test positive

Specificity: Proportion of persons without condition who test negative.

Positive Predictive Value (PPV): Proportion of persons with positive test who have condition

Negative Predictive Value (NPV): Proportion of persons with negative test who do not have condition.

The methods used scored well in specificity, positive predictive value and negative predictive value. What was somewhat lacking is sensitivity–proportion of people who test positive. Many children with ASD’s are not captured by the methods used, suggesting that the estimates by the CDC are likely low.

What would be very interesting to see would be how the sensitivity has varied over time. Is part of the rise in the CDC’s autism prevalence estimates due to changes in sensitivity? Could the records be getting better as a result of the autism awareness campaigns over the last 10 or more years?

7 Responses to “Evaluation of a Records-Review Surveillance System Used to Determine the Prevalence of Autism Spectrum Disorders.”

  1. Kent June 29, 2010 at 00:13 #

    Can one opt out of this surveillance? No? Then this could be dangerous to future generations of autistic people if a fascist government came to power. You should stop and think about this. Also, how will insurance and employers treat this data should they ever have access to it. A little less time spent on arguing over causation on this blog and a little more time thinking about autistic people would be helpful.

    • Kev June 29, 2010 at 17:05 #

      Kent – how about starting a blog of your own to discuss what you think is important about autism? That would certainly be more productive than commenting on other blogs to tell us what we should choose to write about.

  2. Joseph June 29, 2010 at 02:04 #

    In a binary classifier, specificity is nothing more than the true-negative rate, and sensitivity is the true-positive rate.

    So a sensitivity of 0.60 means that if 100 autistic children are classified, only around 60 will test positive.

    40 will give false negatives. Clearly, the false-negative rate is 1 minus the sensitivity.

    The false-positive rate is very low, which is to be expected. It’s like I always say: Overdiagnosis is not the main issue in autism epidemiology. Past underdiagnosis is.

  3. livsparents June 29, 2010 at 02:29 #

    If a facsist government came to power Kent, the ADDM data would be the absolute least of your worries. Just a click into your doctors’, your pharmacies’…heck, just an evaluation of the hard drive on your computer would yield 10 times more damnable data on you if a government were able to tear up the the Bill of Rights.

    The only way to opt out of surveillance is to opt out of services. But at the ADDM level the data is pretty anonymous, it’s down at the state and local levels that it becomes more questionable. NJ is even asking for ‘voluntary’ registration into and ‘autism’ database. But your privacy’s still safe…IF you trust your government. At this point, I’d sooner trust them that even ATTEMPT to keep my personal data completely private…

  4. Kent June 30, 2010 at 00:32 #

    Kev asks why I don’t start my own blog to discuss issues related to autism.

    Kev, I spend my advocacy time changing lives in my local community. I don’t have the time to blog. I spend my advocacy time as a board member of the state’s mental health board advocating for autistic people in my state.

    Is blogging the only legitimate avenue to comment on autism in your view?

    @livparents, I don’t trust database collections or citizen surveillance by my government, regardless of party.

  5. Kev June 30, 2010 at 08:08 #

    No Kent its not. But your comment touched on this *blog* hence my answer touched on *blogging*

    If you have time to comment on numerous blogs, you certainly have time to start your own blog. I manage both off and online activities for autism,manic depression etc and am a family man. I’m not suggesting you give over your life to blogging, merely that you stop telling others how to conduct their advocacy efforts.


  1. Tweets that mention Autism Blog - Evaluation of a Records-Review Surveillance System Used to Determine the Prevalence of Autism Spectrum Disorders. « Left Brain/Right Brain -- Topsy.com - June 28, 2010

    […] This post was mentioned on Twitter by Kev, Edgar H Phillips. Edgar H Phillips said: Could autism prevalence be higher than we think? http://bit.ly/9HEmXP […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: