Saturday, June 11, 2011

LA times Korean Autism study.,0,7819598.story

First there was a write up on the study.  One of the posts in the comments was particularly interesting to me and I wanted to save it. 

mudfud1 at 10:37 AM June 8, 2011
I'm a physician-scientist and I've read the actual study manuscript as well as the papers that developed the questionnaires and tools used to diagnose autism.

The study has been designed and conducted quite well, but the conclusions that have been drawn are fundamentally flawed. To be sure the medical literate is littered with with flawed studies, but I'm troubled by the lack of critical thinking and analysis regarding this very important study; one that could have major public health implications.

As as example, if you look at the receiver operating characteristic (ROC) of the diagnostic questionnaire, it will show a high false positive rate when applied as a screening tool for a general populations (as opposed to a targeted population). The positive predictive value would be around 20% (the actual value would depend on the true prevalence), which means that 80% of the children diagnosed as "autism" are false positives.

It is unfortunate that such important points are not considered at all, and the data and conclusions are simply accepted as fact. The negative implication are not just for the kids incorrectly diagnosed as being autistic, but also for the children who truly have autism, because resources and funding will be diluted for the children who really need help.

No comments: