Editorial: Evaluating the Success of Early Detection of Autism: It's Time to Move Beyond the Median.

For many years, the median age of diagnosis has been the most widely used metric for evaluating changes in the early diagnosis of autism. The logic appears simple: if health and educational systems improve their ability to diagnose autism at early ages, then it would seem that the median age should fall. However, despite decades of efforts in the United States to improve the early identification, the median age of diagnosis has hardly budged. But simple logic can be deceptive. In a watershed analysis of data from the Autism and Developmental Disabilities Monitoring Network-the most prominent study of autism prevalence in the United States-Shaw and colleagues apply different metrics to existing data to yield novel findings. Citing earlier methodological work published in this journal and others, the authors report what they refer to as the "simple median age of diagnosis" (ie, the median age at the earliest recorded autism spectrum disorder diagnosis or special educational eligibility before 8 years of age) alongside 2 alternative metrics: (1) "median age at identification"-similar to the "simple median" except that it also includes children who meet case definitions for autism but are not yet identified by 8 years of age (ie, they do not have a recorded diagnosis or record of special education eligibility); and (2) cumulative incidence of autism by 4 years of age-ie, the ratio of the number of children diagnosed by 48 months of age divided by the total number of children in the population. Results have tremendous implications both for research and policy.