A recent study conducted by Dr. Young Shin Kim and colleagues has received a lot of attention over their findings, which estimate the current prevalence of autism spectrum disorders (ASD) in South Korea at approximately 1 in 38 children between the ages 7 and 12 years. As this statistic is a staggeringly greater than the current estimates of ASD prevalence in other parts of the world (e.g., roughly 1 in 110 children in the United States, and about 1 in 100 children in England), I feel it is important to take a closer look at this study.
Dr. Kim and colleagues set out to measure the prevalence of ASD in South Korea using a population-based sample of school-age children. The target population included all children ages 7 to 12 years enrolled in Ilsan elementary schools or enlisted in the Ilsan Disability Registry, approximately 55,266 children. Of the 44 Ilsan elementary schools, 33 agreed to participate in the study leaving a pool of roughly 36,592 children enrolled in mainstream education. All the children identified via the Ilsan Disability Registry, a total of 294 children, were also recruited to participate in the study.
In the first phase of the study, parents were asked to complete the Autism Spectrum Screening Questionnaire (ASSQ) to screen the participants for ASD. Next, comprehensive diagnostic evaluations were offered to all the participants enlisted in the Ilsan Disability Registry, as well as all the participants enrolled in mainstream education who screened positive on the ASSQ. From the available pool of 36,886 children, ASSQs were received for approximately 23,337 children. A total of 1,214 children were offered diagnostic evaluations and of those children, only 286 were able to undergo evaluation. So just to recap, the study started with more than 55,000 children eligible for participation but due to logistics and lack of voluntary participation, only 286 children were actually evaluated for an ASD diagnosis.
Due to the severe drop off in participation from those who were qualified to receive diagnostic evaluations but did not, Dr. Kim and colleagues employed methods to adjust for the missing data. For starters, all the children reported in the Ilsan Disability Registry as having ASD were assumed to have it. As for the children enrolled in mainstream education, likelihood of having ASD was predicted based on the data collected from the participants who did undergo diagnostic evaluations, as well as each child’s ASSQ score, sex, and age. Thus, the 1 in 38 statistic that we see reported is based off of these adjusted data, not the actual observed children.
The results of the study revealed both unadjusted and adjusted ASD prevalence estimates. The unadjusted prevalence of ASD, based on the raw data, was found to be 0.36%, approximately 1 in 278 children. On the other hand, the adjusted prevalence of ASD, based on the weighted data, was estimated at 2.64%, about 1 in 38 children.
The estimated ASD prevalence of 1 in 38 children as determined in this study is not based on the raw data that Dr. Kim and colleagues collected, but rather it is based on the authors’ speculation of what their data would have been if they had been able to evaluate a larger sample. Unfortunately, relatively few parents wanted their child to participate in this study. That is, for some underlying reason (whether it is denial, fear of stigma, confidence that their child does not have an ASD, etc.), the majority of parents decided not to go through with the diagnostic evaluation. In spite of this significant response bias, Dr. Kim and colleagues used data collected from a minority of their participants to make assumptions about the majority, which cannot be done with confidence.
While the results of this study may strengthen ASD awareness in South Korea, which is certainly needed, I urge you all to take this study with the proverbial grain of salt. The danger in declaring such a high prevalence of ASD that is only based on adjusted data is that other estimates that are based upon more solid research methods may be dismissed as a result. Until these findings are replicated in further research, we may be better off regarding Dr. Kim and colleagues’ statistic of 1 in 38 as a likely overestimation of the prevalence of ASD in South Korea.
Kim, Y. S., Leventhal, B. L., Koh, Y. J., Fombonne, E., Laska, E., Lim, E. C.,… Grinker, R. R. (in press). Prevalence of autism spectrum disorders in a total population sample. The American Journal of Psychiatry.dio: 10.1176/appi.ajp.2011.10101532