In 1965, Leo Kanner, the father of autism research, expressed his concerns about overdiagnosis: " ... it became a habit to dilute the original concept of infantile autism by diagnosing it in many disparate conditions which show one or another isolated symptom found as a part feature of the overall syndrome. Almost overnight, the country seemed to be populated by a multitude of autistic children."
There followed a period of standardisation of diagnostic criteria. Psychiatrists worked hard to develop instruments to make diagnosis more objective, and cases of autism became less common and more clearcut. The pendulum, however, is now swinging back.
Earlier this month, a press release from Yale University announced: "Prevalence of Autism in South Korea Estimated at 1 in 38 Children". This estimate was considerably higher than recent UK studies (Baird et al, and Baron-Cohen et al). There are three possibilities: either children in South Korea are at unusually high risk of autism; previous studies have dramatically underestimated the prevalence of autism; or this new study gives an overestimate.
She suggests that we cannot dismiss the third possibility.
The South Korean study was not restricted to classic autism, but included the milder condition of ASD. To assess this, the authors used culturally appropriate versions of diagnostic interviews and observational instruments. Expert clinicians then reached a consensus diagnosis on the basis of information from these.
As I have discussed elsewhere, both interviews and observations are often conducted because the conclusions of either – when taken alone – are not watertight. For instance, the observational instrument only assesses how a child behaves during a 40-minute period, and so may miss behaviours that are rare or that occur in less structured settings.
These assessments can also over-diagnose autism, however. For instance, it is possible for children to appear impaired on the domains of social interaction and communication if they are shy, anxious, sullen, or have language difficulties. For all these reasons, it is seen as dangerous simply to rely on numerical scores to make a diagnosis.