The answer is straightforward. No. One may not come to this conclusion, however, when he reads popular media, especially articles written in the US press. The possibility that the rate of autism is increasing has sparked and then fueled a debate on the purported epidemic of this disorder and its causes (1). According to Russell, Kelly, and Golding (2) many lay people believe that increased incidence of autism is due to exposure to new environmental, medical and technological hazards, including vaccinations, folate supplements, pluripotent mast cells, and growth of cable television. However, no sound scientific evidence indicates that the increase in the number of diagnosed cases of autism arises from environmental causes that were not present in the world 70 years ago, but came to being during the last decade of the previous century (3). In this article I will: 1) review the range of prevalence estimates for autism and related disorders, 2) examine the hypothesis of a secular increase in rates of autism, and 3) discuss several major sources of misunderstanding related to the increased number of diagnosed cases of autism.

Information on the rates of autism comes from epidemiological studies, which started in the mid-1960s in England (4) and have since been conducted in many countries (5). These surveys have focused on a categorical-diagnostic approach to autism that has relied over the years on different sets of criteria. In general, they were concerned with the autistic disorder/childhood autism as characterized by severe impairments in communication and language, social interactions, as well as play (6). In other words, the majority of those studies did not investigate other pervasive developmental disorders (PDD). Some recent epidemiological surveys also included PDDs that fell short of strict diagnostic criteria for autistic disorder (i.e., PDD not otherwise specified – PDD-NOS and Asperger syndrome). Another aspect of the epidemiological research is whether it reports the prevalence or the incidence of a phenomenon. Prevalence refers to the number of individuals in a specified population who have the condition being studied at a specified time. It is the number of cases affected with a given condition divided by the population. It usually is expressed as percentage or the number of cases per 1.000 or 10.000 (7). Incidence refers to the number of individuals in a specified population in whom the condition being studied begins within a specified time period, such as one year. Both indexes are important but the usefulness of each depends on the condition being studied. In the case of autism, knowing the incidence would be vital because incidence rates are a more sensitive indicator than prevalence rates of potential new etiological factors (1). Unfortunately, calculating incidence rates for autism is problematic and thus most epidemiological research reports prevalence. Review of the range of prevalence estimates for autism and other PDDs comes from a series of publications by Fombonne (8-12). In the last article, Fombonne reported the most up-to-date review of published epidemiological surveys of PDDs. He included 53 studies published between 1966 and 2008, 43 of which provided information on the autistic disorder and 19 on all PDDs combined. Surveys were conducted in 17 countries and more than 50% of results were published after the year 2000. Most studies were conducted in urban areas, with school-age samples (median age of 8.0) of a mostly non-immigrant status and median population surveyed of over 63 000 subjects, however the variation in the size of the population surveyed was significant. Most investigations relied on a two-stage approach to identify cases in underlying populations and used batteries of standardized measures, specifically the gold standard diagnostic tools such as the Autism Diagnostic Interview – Revised (ADI-R) or the Autism Diagnostic Observational Schedule (ADOS). The diagnostic criteria used to define caseness varied and reflected historical changes in classification systems. The following criteria were used: Kanner’s, Lotter’s, and Rutter’s for studies done before 1980 and DSM- or ICD-based criteria for studies conducted thereafter.

Autistic disorder (AD). Data on children with autism were available in 43 studies. The earliest one was done by Lotter in 1966 in the UK, Middlesex area. The target population was 78 000 children between ages 8 and 10, 32 subjects with autism were identified with the use of a rating scale, about 15% of them had IQ above 70, gender ratio of males to females was 2.6:1. The estimated prevalence was 4.1/10 000. The latest one was done by Latif and Williams in 2007 also in the UK, Wales region. The target population was slightly over 39 000 children between ages 0 and 17, 50 subjects with autism were identified with the use of Kanner’s criteria, no information was reported on the IQ score, neither on the gender ratio of males to females. The estimated prevalence was 12.7/10 000. For all 43 surveys, prevalence estimates ranged from 0.7/10 000 to 72.6/10 000. Prevalence was negatively correlated with sample size, that is the smaller sample the higher prevalence. The correlation between prevalence and year of publication was significant. All studies published after 1987 reported prevalence higher than 7/10 000. In 18 studies published since 2000, the prevalence varied from 7.2 to 40.5/10 000. The average prevalence for the autistic disorder was 20.6/10 000. The author writes that this value can be used as “the best current estimate” for the prevalence of AD (12), str. 592.

PDD-NOS/Unspecified PDD. Seventeen of the 43 provided separate estimates for the prevalence of the atypical autistic syndromes, which were labeled differently but corresponded to the current diagnostic criteria for PDD-NOS. Fourteen of those 17 studies showed higher prevalence for those disorders than for autism. The ratio of the prevalence for PDD-NOS to the prevalence of autism had a mean value of 1.8, which translates into an average prevalence estimate of 37.1/10 000 for PDD-NOS if we take the value of 20.6/10 000 as the prevalence for autism.

Asperger Syndrome. There are very few epidemiological studies of Asperger Syndrome (AS), most likely due to the fact that this disorder was acknowledged as a separate diagnostic category only in the early 1990s with the revisions of ICD and DSM (ICD-10 and DSM-IV). Only two epidemiological surveys focused exclusively on the prevalence of AS. Unfortunately, due to extremely small number of cases identified (fewer than 5), the resulting prevalence estimates are imprecise. However, from the recent autism surveys we can gather information on the AS prevalence. It is consistently lower than that for autism. It is difficult to establish how much lower, but the ratio of 3 or 4 to 1 appears appropriate. This translates into a prevalence rate for AS of 6/10 000. Fombonne, however, cautions that there are strong limitations of data on AS.

Prevalence for combined PDDs. Based on the calculated estimates for each of the PDDs, the prevalence for all PDDs is approximately 63.7/10 000 (i.e., the sum of estimates for AD, PDD-NOS, and AS). Although these results should be treated as estimations, 19 recent epidemiological studies that have focused on PDD as the case definition obtained similar values (specifically 63.5/10 000). Also, surveys done by the Centers for Disease Control in 2002 and 2006 showed the prevalence for combined PDDs to be at least 60/10 000 (13, 14). Fombonne writes that the estimates of around 60 to 70 per 10 000 for all PDDs represent “the best estimate for the prevalence of PDDs currently available” (12, p. 593).

In order to answer the question regarding presence of the epidemic of autism, we must prove that there is a secular increase in rates of autism. Time trends in prevalence and incidence data should be evaluated. Data available from epidemiological research should also be analyzed in the light of such methodological requirements as: constant case definition and constant case ascertainment. Various researchers investigated several factors related to shown increases in the rates of autism (5, 15, 16). First – referral statistics. If we take the number of children referred to special services as a proof for increased incidence of PDDs, we may be mistaken. Increased number of referrals can be confounded by referral patterns, availability of services, public awareness, decreasing age of diagnosis, and changes over time in diagnostic concepts and practices (12). Such confounds can be found in the report from the Department of Developmental Services in California, which is often quoted as evidence for the epidemic of autism (11). Thus, the information provided in the report and showing an increase of several hundred percent in the rates of autism referrals, has to be evaluated critically because the numbers fail to account for changes in the size and composition of the underlying population, no attempt was made to control for changes in diagnostic concepts and definitions, the fact that autistic children are diagnosed nowadays at a much earlier age did not motivate the researchers to get age-specific rates among older children, and finally, changes in the rates of other disorders were not taken into account. Fombonne (12) summarizes that evidence from referral statistics are meager and weak to support the idea of secular increase in rates of autism.

Second – comparisons of cross-sectional epidemiological studies. Many of the epidemiological surveys of autism are characterized by unique design features which makes it difficult to compare their results on the rates of the disorder. Time trends in prevalence and incidence of autism are, thus, very difficult to gauge. If we compare eight surveys conducted at roughly the same time and with similar age groups in the UK and in the USA, four in each country, we will find out large differences in prevalence estimations. As no passage of time was involved, most likely the differences can be attributed to variations in case identification methods (intensive population-based screening techniques vs. administrative methods for case finding). Fombonne writes: “no inference of trends in the incidence of PDDs can be derived from a simple comparison of prevalence estimates over time, since studies conducted at different periods are likely to differ even more with respect to their methodology” (12, p. 595) than studies done at the same time.

Third – repeat surveys in defined geographical areas. Repeated surveys, which use similar methodology and are conducted in the same geographical area at different points in time, can provide useful information on time trends. There were several studies done that strived to achieve this goal, but they also were flawed methodologically. Namely, different age groups were included in the surveys, improved detection among mentally retarded was not controlled for, neither were the changes in the diagnostic concepts and criteria. Thus, prevalence estimates which indeed were increasing should not be taken as proof for an increased incidence in the rate of autism. On the other hand, two surveys (17, 18) which were performed very rigorously did not result in the prevalence rates that were statistically different.

Fourth – successive birth cohorts. If one examined data from well-designed large surveys encompassing a wide age range and found increased prevalence among most recent birth cohorts, this could be interpreted as indicating secular increase in the incidence of a disorder. However, the increased prevalence should be specific only to the disorder of interest. For example, an analysis of special education disability data in Minessota showed a 16-fold increase in the number of children diagnosed with a PDD from 1991-1992 to 2001-2002 (19), but the increase was not specific to autism and also was observed for other disability categories, which may be indicative of better services and methods for diagnosis. Moreover, in the early 1990s PDDs were included in the federal Individual with Disabilities Educational Act (IDEA) funding and reporting mechanisms in the USA.

Fifth – incidence studies. Three studies provided incidence estimates for PDDs. All showed an upward trend in incidence over short periods of time, none of them, however, could determine the impact of other factors (e.g., changes in the diagnostic criteria, improved awareness and service availability) on the upward trend.

To summarize, the available epidemiological studies indeed show an increase in the prevalence of autism, but those estimates cannot be directly attributed to the increase in incidence of the disorder. Rather, other variables may be responsible for such state of affairs. They will be discussed in the next section.