Autism Prevalence Trends Over Time: The Confounding of Changes in Age at Diagnosis

Background: Several studies have shown an increase over the last two decades in the age-specific prevalence of autism. The apparent increase in autism prevalence may be confounded by changes in the age at diagnosis, in studies with insufficient follow-up time to estimate the lifetime cumulative prevalence of autism.

Objectives: Examine if there is evidence for changes in age at diagnosis on the reported diagnosis of autism in Denmark. Further, to estimate the amount of bias in the reported prevalence of autism caused by changes in the age at diagnosis.

Methods: The study cohort consisted of all children born in Denmark from 1994 through 1999 identified in the Danish Medical Birth Register (approximately 407,000 children). Outcome data consisted of both in- and outpatient diagnoses reported to the Danish National Psychiatric Registry from 1995 through 2006 using the ICD-10 diagnostic coding system (N=2649). The study population was divided into 2 year cohorts to examine the changes in age at diagnosis.

Results: The study confirms increases in the prevalence by age of both autism spectrum disorder and childhood autism seen in other studies. There was, however, evidence for changes in the distribution of age at diagnose. Changes in age at diagnosis and differences in follow-up time between the two-year cohorts accounted for 52% of the apparent increase in Autism Spectrum Disorder (ASD) prevalence at age 3 in the 1996-97 cohort, and 66% of the apparent increase in ASD in the 1998-99 cohort, compared to the 1994-95 cohort.

Conclusions: Shifts in age of diagnosis, especially the substantial acceleration at younger ages, inflated the observed prevalence of autism in young children in the more recent cohorts compared to the oldest cohort.