The prevalence and characteristics of autism spectrum disorders in the ALSPAC cohort

Abstract

See related article on page 672 The prevalence and sociodemographic characteristics of children with autism spectrum disorders (ASD) remain an important public concern. Although reports of the prevalence of ASD vary widely,1 it is generally accepted that the prevalence is higher today than in the past and there have been recent reports of prevalence estimates as high as 2.7% of children.2 In countries where special education services are available, determining the prevalence of developmental disabilities, including ASD, from multiple sources is enhanced by the ability to identify children from education records. Prevalence estimates using this methodology have yielded estimates that are similar across sites and are higher when education records are included in addition to health records.3 Williams et al. describe the prevalence and characteristics of 11-year-old children with ASD in the UK in 2003 using a large representative sample of children who have been followed since birth in the Avon Longitudinal Study of Parents and Children (ALSPAC).4 ALSPAC offers a unique advantage: the ability to examine both health and education records in order to identify children with ASD. In the study, ICD-10 coded diagnoses for ASD were used to identify children with a developmental delay; a study diagnosis of ASD was then based on findings from a multidisciplinary assessment, as reviewed by an experienced clinician. A second independent source of cases for the study was from the PLASC (education) dataset, which included records of children in special schools; ASD as a primary or secondary concern was used to identify cases for the study. Almost all of the children in the study (94%) were found to have a statement of special educational needs, but because of privacy concerns, there was no way to link the health and education records in order to get a composite record for each child. But, no matter how complete the ascertainment of cases appears to be from multiple sources, some children might be missed because of under ascertainment of children with milder disabilities. Yet, in this study, the small percentage of children with childhood autism who had intellectual disability (mental retardation; 14.7%) or epilepsy (10.3%), suggests that more severely affected children may have been under ascertained. From ALSPAC, the overall prevalence of children receiving special education services for autism was 61.9 per 10 000 children, about 10 per 10 000 higher than the prevalence obtained from health sources using the ICD-10 diagnostic code for ASD, and consistent with results from other recent population studies.3 It has been documented that casting a wider net, e.g. applying behavioral criteria consistent with ASD from the records of children with other relevant clinical diagnoses, as well as those receiving special education services other than for ‘autism’, will yield more children who meet surveillance criteria for ASD. Data from a large multisite surveillance program in the US found that the prevalence of ASD among children who were ‘previously classified’ (i.e. had a diagnosis of ASD and/or received special education services for ‘autism’) was approximately 1.1 to 3.2 per 1000 children lower compared with the prevalence of ASD casting the wider net.3 The median age of diagnosis was available only from health records, and was 44.9 months (range 37–88mo), comparable to that of 8-year-old children (range 49–66mo) in the US.3 The male: female ratio of 6.8:1 with a trend toward females being more severely affected and the sociodemographic findings related to ethnicity, maternal education, and maternal age were not new findings. However, specific sociodemographic risk factors for ASD have been found to differ according to whether identification of the children with ASD was from education records or from health records. Overall, the findings from this study tell us once again that ASD is a common disorder of childhood and confirms that including education records as part of a multiple source ascertainment strategy yields a higher prevalence of ASD than if only health records are used. The results from ALSPAC indicate that, at a minimum, approximately 60 per 10 000 children in the UK have ASD and, based on several potential sources of under ascertainment of cases, the true prevalence of ASD in this community is likely to be even higher.