Prevalence perturbed

Reports of soaring autism rates in recent years have driven debate over whether the numbers reflect true increases in the incidence of autism. Critics have cited a range of factors that can inflate apparent prevalence, from inconsistency in parents’ responses on surveys, to improved diagnostic services and the many forms of sampling bias.

Many of these factors may be at play in a new study showing a steep rise in autism rates in parts of Canada.

Autism prevalence in three regions of eastern Canada rose by as much as 15 percent in less than a decade, but only part of that increase seems to be genuine, according to a report published 16 June in the Journal of Autism and Developmental Disorders.

The researchers identified several factors, such as new diagnoses of autism in older children and administrative errors in recording families moving in or out of a region, that may have contributed to an apparent increase in the regions’ prevalence.

The work bolsters arguments that prevalence data are prone to inflation, but suggests that careful analysis of the data can also point out weaknesses in the system that serves autism families.

Beginning in 2002, the National Epidemiologic Database for the Study of Autism in Canada gathered basic information such as age, sex, birth date and first local autism diagnosis from children with autism in Canada. The children ranged in age from 2 to 14 years. This report followed populations of children with autism from the province of Newfoundland and Labrador, Prince Edward Island and the six counties that make up Southeastern Ontario.

The researchers tallied the numbers of children diagnosed with autism each year and calculated prevalence estimates for every 10,000 children within three age groups: 2-5, 6-9 and 10-14 years. Data collection extended until 2008 in Newfoundland and Labrador and 2010 in the other regions.

Data about the children came from diagnostic clinics, schools and behavioral intervention programs. The researchers also gathered medical and familial information from interviews with parents.

Over the course of the study, the researchers found the prevalence of autism increased each year by, on average, 14.6 percent, 9.7 percent and 13.8 percent in Newfoundland and Labrador, Prince Edward Island and Southeastern Ontario respectively. But those regions began with a prevalence of 46, 47 and 53 per 10,000 children, well below the estimated 114 per 10,000 children in the U.S., suggesting that these regions are catching up in their diagnoses of autism.

The prevalence among boys remained roughly four times that of girls throughout, in keeping with the findings of the vast majority of other prevalence research.

Heavy reliance on parental reports can skew prevalence data. Nearly half of the parents from Prince Edward Island participated in the interviews, but only a third of parents in the other two regions did. And parents with younger children were less likely to participate in the interviews in all regions. It is unclear how the missing information might have influenced the results.

Some of the factors that contributed to the prevalence increase would do well as cautionary tales. For example, a lack of services for preschool children with mild autism in Southeastern Ontario made them essentially invisible to the system. Prevalence in this area remained unusually steady among 2- to 5 year-olds, but shot up once children entered the 6- to 9-year-old category and received an autism diagnosis. This phenomenon of diagnosing children with autism only after they have aged into a system that identifies them is often termed ‘catch-up’ diagnosis.

In Southeastern Ontario, where the population of children with autism ranged from about 400 to 900 children, certain agencies failed to report some of the families that had moved out of the area. They retained 29 children with autism on the rolls, keeping the apparent prevalence of autism artificially high in that region.

When the person responsible for making most of the diagnostic assessments on Prince Edward Island returned to the island after an absence, prevalence seemed to skyrocket as the person cleared a backlog of cases.

Overall, the researchers recognize that the prevalence estimates in the study do not precisely reflect the true incidence of autism, but they see the analysis as a learning moment to better track and support autism families.