Article Text
PDF
Statistics from Altmetric.com
Question
Question:
What is the prevalence of autism in Cambridgeshire?
Population:
Children aged 5–9 years and their parents from all schools in the English county of Cambridgeshire (including mainstream and special schools in both private and state sectors).
Setting:
Mainstream, private and special schools in Cambridgeshire; surveyed February 2003–March 2004.
Assessment:
A survey of autism spectrum conditions (ASC) was performed using the Special Educational Needs (SEN) register. Schools submitted both the number of children aged 5-9 years at the school and the number of children with ASC recorded on the SEN register. Participating schools were sent questionnaire packs to distribute to the parents of children aged 5–9 years, including a diagnosis survey; information about behaviour or developmental concerns raised by teachers or professionals; the Childhood Autism Spectrum Test (CAST) plus questions about parental education and socioeconomic status. Families reporting a child with a diagnosed ASC were sent another form asking for further details, including the subgroup diagnosis if known. Confirmed cases were children diagnosed by a relevant health professional. These children were excluded from further assessment. All other children scoring 15 or above in CAST plus a randomly selected 33% of children scoring 12–14 were invited for a detailed assessment. Cases were defined using the Autism Diagnostic Interview-Revised and Autism Diagnostic Observation Schedule criteria, plus consensus diagnosis. Prevalence estimates were adjusted for known non-response, and assessment invitations were weighted towards high scorers.
Outcomes:
Prevalence of ASC (ICD-10).
Methods
Design
Cross sectional (survey).
Main results
Of the 162 schools approached, 79 participated and reported 8824 children aged 5–9 years and 83 cases of ASC from the SEN register. Prevalence estimates from the SEN register were 11 per 10 000 (95% CI 5 to 21) for childhood autism and 94 per 10 000 (95% CI 75 to 116; or 1 in 106) for ASC. The response to the diagnosis survey was 29%, giving 3373 questionnaires for analysis (50% boys and 50% girls). Parents reported 41 cases of ASC using the diagnosis survey. Four of these were from special schools and assumed to be confirmed cases. Further information was provided by 24 parents, from which 19 cases were verified. The prevalence estimate from the diagnosis survey was 99 per 10 000 (95% CI 65 to 141, or 1 in 101). Prevalence estimates for boys were 153 per 10 000 (95% CI 94 to 217, or 1 in 66) and for girls 42 per 10 000 (95% CI 9 to 79, or 1 in 208). Ninety children scored 15 or more on CAST and 52 were assessed further (58%). 122 children scored 12–14 and 25 attended further assessments. The directly observed prevalence estimates from the screened population was 22.2 new (undiagnosed) cases. Adding the known cases and the undiagnosed case reveals 55.5 cases. Dividing this by the known cases gave a ratio of 1.67 (95% CI 1.24 to 2.32) as a multiplier to apply to prevalence estimates. Overall, the ratio of known:unknown cases was 3:2. The overall adjusted prevalence estimate of ASC is 157 per 10 000 or 1 in 64 (95% CI 99 to 246) in Cambridgeshire.
Conclusions
Autism spectrum conditions may affect between 1% and 2% of children aged between 5 and 9 years in Cambridgeshire.
Abstracted from
Commentary
This study investigated the prevalence of autism spectrum conditions (ASC) in 5–9 year olds attending school in Cambridgeshire. As in Canada, the USA, Japan and the UK,1 the authors ended up estimating the prevalence of ASC between 1% and 1.6%. The actual estimates for prevalence varied from 47/10 000 to 165/10 000—that is, a 3.5 fold variation in estimates depending on different assumptions.
The authors went a long way to estimate the proportion of undiagnosed cases in order to adjust the final estimate, taking into account the proportion of false negatives that inevitably exist in such surveys. While the authors argued that two undiagnosed cases in the population can be found for every three diagnosed cases, this conclusion was based on unverifiable assumptions. Firstly, response rates were extremely low: many schools did not participate in the study while participating schools failed to provide data on students listed on the Special Education Needs (SEN) register. Participation rates in the parental survey were similarly very low. Secondly, there was uncertainty on the exact denominator and numerator, as some children aged 4 or 10 years were included. Thirdly, methods used to confirm caseness were uneven and relied on unchecked assumptions. Fourthly, the sampling weights used to estimate the population prevalence assumed lack of bias in factors associated with participation, an unlikely hypothesis. Sensitivity analyses nevertheless provided estimates for different hypotheses, consistent with an ASC rate between 0.5 and 1.6 per 1000 children.
A conservative conclusion from this study is to consider a prevalence close to 1%, as derived from the SEN register as well as from the parental survey. This rate agrees with that found in recently published surveys. A minimum prevalence of 1% translates into a high number of children with ASC, and should draw the attention of service providers to the important needs of this often underserved population. As pointed out by the authors, the increasing prevalence reflects the recognition of milder phenotypes and of ASC in children with a normal IQ. However, alternative interpretations should also be entertained, and call for long term monitoring of time trends of ASC in child populations worldwide.
References
Footnotes
-
Sources of funding Shirley Foundation.
Footnotes
-
Competing interests YC is a full time employee of Lilly. NCP has received research support from Takeda Pharmaceuticals North America and the Thrasher Foundation, and is a consultant for Lilly and Shire. The opinions expressed in this manuscript are solely those of the authors.