Objective: To study autism over time in order to ascertain whether there has been an increase in its prevalence in recent years.
Method: All English language papers on the prevalence of autism were reviewed. Ten of the studies retrieved were not used in the final analysis because they did not meet full criteria for inclusion in the review. The remaining 20 studies, published between 1966 and 1997, were subdivided into US studies (n=2) and non-US studies (n=18), and the latter group was subdivided into four 8-year periods.
Results: The early studies yielded prevalence rates of under 0.5 in 1000 children, whereas the later ones showed a mean rate of about 1 in 1000. There was a marked difference in prevalence rates between those studies that included some children born before 1970 (low rates) and those that included only children born in 1970 and after (high rates). This is probably due to the lower rates obtained by use of criteria strictly based on Kanner's description of his syndrome. The US studies reported atypically low rates. There was a highly significant estimated increase with calendar year in the non-US studies (3.8% per year).
Conclusion: It is concluded that autism is considerably more common than was previously believed. The possible reasons for the higher reported rates are discussed.