From Questioning Answers
By Paul Whitely
October 17, 2013
A short post today based on the paper by Brent Taylor and colleagues* (open-access) adding to the considerable literature on the autism numbers game. The headlines generated from this study are no better encapsulated than that of the BBC: "UK autism cases have 'leveled off'".
The crux of the paper is that based on an analysis of the UK General Practice Research Database (GPRD) which carries details of several million patient records following patient contact with what's known as a General Physician (GP) here in the UK, cases of autism spectrum disorder included in those records were counted and annual prevalence and incidence rates based on 8-year olds were generated.
The results: cases of autism recorded on the GPRD suggested that "the annual prevalence of autistic spectrum disorders was estimated at 3.8 per 1,000 boys and 0.8 per 1,000 for girls".
Importantly, from the BBC: "The study concluded there was "compelling evidence that a major rise in incidence rates of autism, recorded in general practice, occurred in the decade of the 1990s but reached a plateau shortly after 2000 and has remained steady through 2010"".
There are also a few interesting nuggets of information to take from the Taylor paper such as the reason(s) put forward for the dramatic increase in cases witnessed during the 1990s. Another quote I'm afraid on whether greater awareness or broadening diagnostic criteria or diagnostic substitution were the sole causes of the increase:
"It seems unlikely that these factors materially explain the extraordinary increase in the number of children diagnosed in the 1990s; nor the steady state that followed thereafter in 2004 through 2010".
This is in line with what other commentators have talked about (see here).
The Taylor paper is an interesting one and no doubt will generate some discussion about the numbers of cases of autism. I note that the figures reported by Taylor and colleagues are somewhat at odds with other studies using different data collection methods based in other parts of the world such as that 1 in 50 figure in the US discussed quite recently (see here) or other incidence data (see here).
Whether this is down to how the data are collected and verified or truly representative of differing rates of autism in different geographical populations is yet more substance for discussion.