Abstract
BACKGROUND: Since Kanner's first description of autism there have been a number of changes in approaches to diagnosis with certain key continuities . Since the Fourth edition of the Diagnostic and Statistical Manual (DSM-IV) appeared in 1994 there has been an explosion in research publications. The advent of changes in DSM-5 presents some important moves forward as well as some potential challenges. METHODS: The various relevant studies are summarized. RESULTS: If research diagnostic instruments are available, many (but not all) cases with a DSM-IV diagnosis of autism continue to have this diagnosis. The overall efficiency of this system falls if only one source of information is available and, particularly, if the criteria are used outside the research context. The impact is probably greatest among the most cognitively able cases and those with less classic autism presentations. CONCLUSIONS: Significant discontinuities in diagnostic practice raise significant problems for both research and clinical services. For DSM-5, the impact of these changes remains unclear.