In my last post I argued that the apparent marked increase in the frequency of Autistic Disorder over the last three decades is almost entirely due to three factors: broadening diagnostic criteria, younger age at diagnosis, and improved efficiency of case ascertainment. Some autism activists have argued that this can be seen merely as a hypothesis, since models incorporating these three phenomena seem to fit the data from time trend studies quite well, but modeling isn’t a direct test of the hypothesis. They have argued further that a more direct test of the hypothesis woud be to look at a group of adults diagnosed with other conditions related to autism (but not autism) before the current “epidemic.” How would these adults be diagnosed now, if they were tested thoroughly?
A small step in this direction was made recently in an article entitled “Autism and diagnostic substitution: evidence from a study of adults with a history of developmental language disorder,” in the May 2008 issue of Developmental Medicine & Child Neurology. Professor Dorothy Bishop and three colleagues from the University of Oxford’s Department of Experimental Psychology investigated 20 adults who had been diagnosed as having pragmatic language impairment (PLI) as children and 18 adults who had been diagnosed has having specific language impairment (SLI). I don’t want to get bogged down in definitions here, but there are good explanations of these diagnoses at the Help for Kid’s Speech website, which also has links to other sources of information on developmental language disorders. Basically, SLI is characterized by normal social skills but a difficulty in learning language. PLI differs from SLI in that social skills are diminished. Note that none of these 38 had ever been diagnosed with autism. In this study all 38 were given the Autism Diagnostic Observation Schedule - Generic (ADOS-G). The parents completed the Autism Diagnostic Interview - Revised, which asked them to recall their child’s behavior at ages four and five. A current diagnosis of autism was confirmed when both instruments resulted in scores above certain threshold values. The current criteria were expanded to include any participant who scored above threshold values for autism spectrum disorder (ASD) or autism on both instuments; in these cases the current diagnosis was ASD.
Here are the results, which I have recast in a table that’s more understandable to me (and I hope to you). You won’t find this table in the paper, which is available free online (not just as a summary).
Original Diagnosis (as children)
|Current Diagnosis||Pragmatic Language Impairment||Specific Language Impairment||Total|
|Autism||8 (40%)||0 ( 0%)||8 (21%)|
|Autism Spectrum Disorder||3 (15%)||2 (11%)||5 (13%)|
|Diagnosis Unaffected||9 (49%)||16 (89%)||25 (66%)|
|Total||20 (100%)||18 (100%)||38(100%)|
It’s obvious immediately that this study has two flaws, which the authors readily admit. First, it had an extremely small sample size. Second, it only applies to one small class of diagnostic substitutions — those pertaining to developmental language disorders. Professor Bishop and her co-authors have also argued that caution should be used to interpret the results because the parents’ recall of their children’s past behavior could be affected by time or by information on autism they’ve picked up. Absolutely true. On the other hand, recall bias will always be a problem in these kinds of studies, if parents are asked to fill out questionnnaires about their adult children’s childhood behaviors. Indeed, given the recall bias problem, it might be better to forget about the parental questionnaires and stick to diagnostic measures like the ADOS-G.
In a commentary in the same issue of Developmental Medicine and Child Neurology, Professor Patricia Howlin of the Institute of Psychiatry, King’s College, London, concludes that: “This succinct study adds to arguments against the view that incidence of autism has increased over recent decades, and suggests that changes in diagnostic criteria are the most likely reason for the rise in the number of cases diagnosed.” I couldn’t have said it better. However, this small study is only a first step and we need more studies with larger sample sizes.
Next post: Let’s not forget that Bernadine Healy is a Celebrity Doctor — not a scientist.