Commentary: On King and Bearman

Much has been speculated about the origin of increased numbers of children receiving a diagnosis of autism spectrum disorder (ASD) in the last 20 years. This phenomenon has been observed worldwide, in countries where repeated epidemiological surveys or surveillance systems could capture these trends upwards. Interestingly, this acceleration occurred at around the same time in the late 1980s or early 1990s. As these trends were recorded in countries as far apart as the USA, the UK, Denmark, Sweden and Japan, it made it less likely that increases were due to exposure to environmental risk factors that would operate simultaneously in such disparate and contrasting settings. Observers also noted that changes started to occur in the 1980s (the term ‘pervasive developmental disorder’ was used for the first time in 1980, in the DSM-III nosography), at a time when the conceptualization of autism was broadened, more ‘high-functioning’ children with good language and intellectual skills were recognized, and when the view of autism as a severe, qualitatively deviant, disorder was progressively replaced by a continuum of combinations of more or less severe deficits. A new dimensional view of the ASD phenotype has emerged, the boundaries of which with other developmental problems or psychopathological syndromes, and with normal development, have become progressively more difficult to establish. Clinical practice changed with increasing demands for standardization of clinical evaluations embodied by the development of semi-structured diagnostic tools such as the Autism Diagnostic Interview (ADI) or the Autism Diagnostic Observational Schedule (ADOS). Diagnostic changes or substitution were among factors incriminated to account for increased prevalence rates. This phenomenon is not new in medicine, and it made perfect sense to postulate that, when autism became increasingly recognized with corollary improvements in funding and educational policies, a flow from previous diagnoses such as mental retardation (MR) to the new broadened concept of ASD would be observed. New studies suggestive of efficacy of early intensive behavioural intervention added to this momentum.1 Practitioners' and consumers' views changed as developmental trajectories of young children diagnosed with ASD were no longer equated with fixed, lifelong, deficits that could not be overcome. The introduction of the 1990 Individual Disabilities Educational Act (IDEA) in the USA was followed by diagnostic practice changes,2 whereby children previously diagnosed with MR were now diagnosed with ASD, either with (accretion) or without (substitution) a co-occurring diagnosis of MR. Most of previous analyses relied, however, on ecological analyses whereby aggregated data on the prevalence of MR and ASD were compared over time. Ecological studies have well-known limitations, the most important ones being that individual ‘exposure’ (in this case, diagnosis) status is unknown and that confounding factors are notoriously difficult to control for. King and Bearman's3 study avoids these pitfalls by revisiting trends in the California public service using individual diagnostic data on children born before 1987 and observed through a long enough period so that changes in their diagnostic status could be documented at the individual level. The results are impressive in confirming that the phenomenon of diagnostic substitution or accretion accounts for ∼26% of the increase in the caseload of autism in the California public system of services. Further evidence of the validity of their results is shown by the fact that rates of diagnostic substitutions peaked at specific periods of changes in diagnostic practices. The diagrams displayed in figure 3 of their article are particularly eloquent in showing that trends over time in disorders that have different base rates, differential rates of increase and evolving co-occurring patterns can make the interpretation of secular changes very complex. As recognized by King and Berman, the findings apply only to the pathway of substitution from MR to autism. Other pathways have been documented using individual data as well. For example, Bishop et al.4 showed that amongst 38 children initially diagnosed with developmental language disorders, up to 66% met criteria for ASD on either the ADI or the ADOS when re-evaluated as young adults with autism-specific instruments. The rate of diagnostic substitution from language disorder to ASD was extremely high (95%) in those children initially diagnosed with pragmatic language impairment, a feature that characterizes language functioning in older or high-functioning ASD subjects. History is full of examples of ASD children having been (mis)diagnosed with other labels. In decades following Kanner's5 descriptions, multiple labels (symbiotic psychosis, early childhood psychosis, infantile psychosis, early childhood schizophrenia, etc.) were employed to characterize a phenotype that would now fall onto the autism spectrum. Because the phenomenology of autism is largely made of symptoms that are non-specific and often observed in other psychiatric disorders, it is highly plausible that high-functioning autistic children have been diagnosed (and unfortunately are still diagnosed in some countries) with other psychiatric labels such as attachment disorder, obsessive compulsive disorders, sensory integration disorder, learning disabilities (non-verbal) or atypical personalities,6,7 often depending upon the professional background of the evaluator (language disorder for speech pathologists, psychiatric disorder for child psychiatrists, learning disorder for educational psychologists, etc.). By design, King and Bearman could not document these other pathways of diagnostic substitution, which are likely to account for an additional high proportion of the rise of ASDs, especially when not associated with MR. It remains difficult to establish whether or not all the increase seen in many states or countries can be attributed to diagnostic substitution combined with increased awareness. King and Bearman's study, however, makes it clear that a substantial proportion of the increase can be attributed to these changes in diagnostic practices. The implication is that the search for environmental risk mechanism that may have contributed, in addition to changes in diagnostic practices, to the rise in number of children diagnosed with ASD cannot be performed using time trend analyses of datasets or of national registers. The ability to control for the effect of diagnostic changes in these datasets is often limited and, as a result, exploration of environmental risk factors should rely on other research designs such as prospective cohort or population-based case–control studies. Conflict of interest: Dr Fombonne is an expert witness for vaccine manufacturers and for the US Department of Justice and US Department of Health and Social Services in the US thimerosal litigation. None of his research has ever been funded by the industry.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.