We are looking for comments and we are interested in your experiences. Please use this Blog to post your comments.



     Related Articles

Article: increase in autism

The possibility of an association between the MMR and autism has generated much discussion and not an insignificant amount of time investigating this possibility. Most data has ultimately rejected this association, but it was almost entirely been focused on the measles possibility. While attention has addressed the issue of an autism- MMR association, it has been based on the erroneous assumption that the MMR has been unchanged since its introduction in 1969. Overlooked was the fact that the rubella portion was changed from the HPV-77 or Cendehill versions in the original MMR (MMoR) to the rubella 27/3 strain (MMR27/3), introduced in 1979 in the US and subsequently in different years in other parts of the world. A review and focused analysis of the literature, suggests that this is very possibly a significant factor in the now uncontested increased prevalence in autism disorder (AD) and pervasive developmental disability (PDD). (The use of MMR will be used when referring to instances in the literature which do not distinguish between MMoR and MMR27/3.) All prevalence rates are expressed as “x”/10,000 children (unless otherwise noted).

Virtually all articles which address the issue of a possible association between the measles component of the MMR and autism/autism spectrum disorder (AD/ASD) begin with a reference to Wakefield’s article(1). The Wakefield hypothesis emphasizing the short time lapse between the MMR and autistic symptoms as well as GI symptoms and pathology has been generally rejected based on three findings:

  1. no increased incidence of GI symptoms was substantiated,
  2. no difference in the rate of regression with AD has been substantiated, and
  3. no increased rate of AD has been demonstrated to coincide with the introduction of the MMR.
The first finding will not be addressed. The second finding has been questioned for a variety of reasons, not the least of which is that retrospective history is scientifically inadequate to confidently establish an association. The third finding has had a number of problems including not only internal problems which will be addressed below, but also an assumption which turns out to be only partially true. “Increased reporting of ASD in recent years has occurred long after the introduction of MMR vaccine …in 1971 and widespread use of the vaccine in the 1970’s for routine immunization of children at 12 to 15 months of age.” (2) That the measles strain remained unchanged was a valid assumption in analyzing Wakefield’s hypothesis. Unfortunately, this assumption of immutability has unfortunately overlooked the fact that the MMR did not remain unchanged. The rubella strain was initially the HPV-79 or Cendehill strains beginning with the MMoR introduction in 1971 until 1979 when the RA27/3 strain was substituted. (3) This has major implications for a total rejection of the possibility of any association of AD/ASD with the MMR. It is time we reexamine the data for a possible link to the rubella RA 27/3 strain of the MMR27/3. We will first examine the studies used to reject the Wakefield hypothesis, using the change from MMoR to MMR27/3 as a new focus point to reinterpret the findings. Then we will examine the data from countries when and where it is possible to compare AD/ASD prevalence rates pre and post MMoR modification.

WAKEFIELD HYPOTHESIS

            A more review of the issue by Wilson et al (4) addresses four hypotheses: 1) increased rate post MMR vs. no MMR receipt, 2) increased rate post MMR, 3) temporal association with MMR and 4) “new variant form” post MMR.  It must be remembered that this body of literature addressed the Wakefield hypothesis of an association between the MMR (both older MMoR and newer MMR27/3) with both autism and GI symptoms. Much data addressed the association between the MMR and bowel symptoms, but will not be addressed in this review.           

           Increased rate in MMR vs. no MMR          

          To compare vaccinated with non-vaccinated, Madsen et al (5) looked at children born between January 1, 1991 through December 31, 1998 but the collection of cases was stopped on December 31, 1999  thereby leaving out AD/ASD cases which were discovered  later. Not only does this cohort fail to follow up for a minimum of 2 years post- MMR vaccination, some could not even have received their MMR vaccine, being less than 12 months age at the end of the study. Moreover, the null group included months in which children temporarily were delayed in receiving the MMR, counting these months without the vaccine as time supporting no association. The former artificially underestimates the rate of AD/ASD and the latter two artificially  inflate the null association.  The premature termination of the study invalidates the results. (see below for further analysis)          

          Increase rate post MMR
         The MMR27/3 was introduced in England in October, 1988. Taylor et al (6) studied children born from 1979 - 1992 evaluating data in mid 1998. While they recognized the existence of “a significant upward trend over the period 1979-92 for core and atypical cases…and a nearly significant upward trend for Asperger’s syndrome…there was no evidence of  sudden step up in (the) 1987” group, being the first to receive the MMR. (6) Moreover, there was also a recognition of  “higher functioning children with autism and Asperger’s syndrome“ (6) suggesting something new happening.  If the MMR27/3 was a cause, they claimed that there should have been a reasonably parallel “step up” in autism after the 1987 group, which did not occur. Excluding the possibility of the rubella RA27/3 vaccine being used prior to 1988, the reality of retroactive “catch up” immunizations in children not immunized prior to 1988 would explain the “upward trend” documented back to earlier birth cohorts (BrCo’s). In addition, “an increasing number of professionals trained to recognize the disorders” (6) was noted which could also contribute to the gradual, rather than sudden upswing.  A later article by the same lead author looked at a larger group spanning “20 years from 1979” looking for an increased rate of developmental regression. (7) Without addressing the difficulties of a diagnosis based on retrospective history, the study had the same weaknesses as the first, failing to address the existence of retroactively given catch up immunizations.
         This brings up the dilemma of differentiating the rate of AD/ASD in its classical form from newer cases caused by MMR, which was claimed to be reflected in a different rate of regression and a short latency. But disease incidence and rater of regression are not the same. This has added to the confusion, which was addressed by Fombonne & Chakrabarti (8) in a comparison of three groups: the MFS sample with a birth cohart (BrCo) 1954-79, resulting in 99 cases over 25 years, the MHC sample with BrCo 1987-96, resulting in 68 cases over 9 years, and the Stafford sample, with BrCo 1992-95, resulting in 97 cases over 4 years. The value of this study firstly, lay in “rul(ing) out a precipitous onset (of AD) within days after immunization as reported” (8) by Wakefield et al. (1). Secondly, finding no change in (probable, definite or any) regression, it was concluded that “there was no evidence that regressive autism has increased in frequency.” (8) Moreover, the AAP conference reported on the lack of “standard definition of what is meant by regression.” (2) Too often this is dependant on parental recall, helpful but not entirely reproducible. In the final discussion, they admitted that the lack of change in rate of regression “does not rule out this possibility” of “separate causal mechanisms.” (8) (See below for further discussion.)
         Two Swedish articles take some interesting turns in rejecting the AD/ASD-MMR association. In the review (9) of the 1991 Swedish article (10) using a 1975-84 BrCo, it is suggested that “the apparent increase is in part due to better detection, but also to new cases born to immigrant parents.” (10) The concluding paragraph notes that the autism rate was “about double that previously reported”  (10)(even without the inclusion of Asperger’s syndrome). The addition of Asperger’s syndrome would have enhanced the doubling to “several times higher” (10) for AD/ASD. The 1998 report was only a commentary without a detailed breakdown of the data. While claiming “no support for a connection“ (9), there were two BrCo’s, pre- and post 1982 introduction of the MMR27/3 in Sweden. Since there was no enhancement of autism numbers after the MMR 27/3 was introduced, it was taken to prove no association. (9) Viewed strictly from the MMR27/3 use, the authors report only a gradual increase in coverage for measles to about 50% pre introduction of the MMR27/3; but after introduction, the coverage rate rapidly reaches 90+% (9), implying that not only the post-MMR 27/3 BrCo, but that also at least a portion of the previously un- immunized pre-MMR27/3 cohort received a catch-up vaccination. This would mean that potentially 45% (50% un- immunized x 90% success) of this  pre-MMR27/3 cohort was actually immunized with MMR27/3 as well. This lack of clarity plus a lack of statistical testing, make the conclusion excluding an MMR27/3-AD/ASD association a weak conclusion.
          The next two studies are important in establishing a definite increase in AD/ASD. Both collect data only after the introduction of the MMR27/3 - from 1988 -93 in England (11) and from 1980-94 in California. (12) A “nearly fourfold” increase in the incidence of AD/ASD from 1988 thru 1993 (11) can be roughly duplicated in the US chart over the 1987-1994 span. (12) Both have a relatively flat MMR immunization rate which does not coincide with the increased rates noted with AD/ASD. This dissonance in curves is taken as a rejection of the MMR- AD/ASD connection. While it was recognized that an “increased awareness…among parents and general practitioners” (11) could have explained the dissonance in curves, neither attempted to dissect out this factor. Moreover, the availability and utilization of a new MMR 27/3 vaccine in the US was a gradual process as the older versions of vaccines (MMoR) were used up. (Given that vaccines usually have at least a two year expiration date, the gradual penetration of the immunization schedule could be expected to take at least two years.) This should not be ignored in the rejection of an MMR-AD/ASD association.           The fact that four studies attempt to deny an increased rate while two other studies clearly document an increase in rate of AD/ASD reflects an internal conflict in data which the Wilson review  fails to address.
                          Temporal Association
          The difficulty of excluding a connection based on a temporal association is misleading if the follow up is excessively short. But some insight can be gained. Patja et al (13) is clearly limited by recording problems exclusively within 18 days post vaccination, based on the natural rubella disease incubation period.  DeWilde et al (14) showed only that there was no “difference in numbers of consultations” from six months before compared to six months after MMR receipt.  Given that many cases of autism are not found until 12 to 24 months afterwards, this is a relatively irrelevant study. The failure to find any cases within these time frames hardly exclude the MMR27/3- AD/ASD association.
             Madsen et al (5) claim to show “strong evidence against” the association, but the whole issue of  completeness in diagnosis was short changed by obtaining data received only through the end of 1999, a mere 12 months after  the last birth January 1991 to December 1998. This means that approximately 11% in the entire group would not have received the vaccine because of  age and conservatively 40% of AD/ASD cases would have been missed, being too young, given the mean ages of diagnosis at 4.25 years/5.25 years (5).
              Fombonne and Chakrabarti (8) have lots of data but their focus is on the incidence of rate of regression, since Wakefield claimed that the association of MMR and autism was proven by an increased incidence in the rate of regression between the pre- and post-MMR groups. Failure to detect a difference in the rate of regression helps denigrate the Wakefield hypothesis but does not ipso facto eliminate any association.  As they said, “more precise and replicable…data” is needed before adding to “ambiguity and confusion that is damaging to both the public health and the science.” (8)
           Three studies in the Wilson review were presented as studying age of ASD diagnosis & age of first parental concern in MMR vaccinated vs. non vaccinated . While of some interest, the whole area is rightly criticized as being distorted by retrospective memory, a notoriously inadequate scientific measure.  While this needed to be explored, it neither supports nor disproves an association.
           Farrington et al (15) addressed the question raised regarding the limited nature of the case series method used to evaluate a temporal association, but which was valid only for measuring outcomes within short time periods after vaccination. (16) Farrington et al reanalyzed earlier data and concluded that “the relative incidence is not significantly different from 1, indicating no association” (15) but there are problems. Table 1 indicates that the relative incidence goes from 1.06 for MMR vaccine(s) to 2.03 for “any measles-containing vaccine(s)” (i.e. MR - presumably MR27/3) (15) with no discussion of why the doubled rate, which seems to contradict the conclusion of no association.  It is also noted that the ‘99 article found the “relative incidence for atypical cases when analyzed separately remained raised at 1.99” with no elucidation of the doubled rate. (6) This is especially confusing when the increased number of AD and ASD cases was found by adding “measles, mumps and rubella vaccine(s)“data in Table 2. (6) The discrepancies may mean less for a measles-autism connection but may be more meaningful for a MMR27/3-autism connection.
            Makela et al. (17) rejected the MMR-autism association based on a lack of “significant clustering”  without showing a comparison of data between vaccinated and unvaccinated. The single Figure 3 demonstrates a “relatively steady (number of admissions) during the first 3 years and then gradually decreased.” (17) The steady rate distributed over three years was felt to reject the MMR- autism association with a short latency proposed by Wakefield. The existence of an “undefined latency“ was noted. Lacking sufficient numbers of non-immunized patients, there no comparison of non-vaccinated vs. vaccinated progressing to autism. Clustering would support an association but only comparison data would negate an association. A prolonged latency period does not rule out an association.
            More recent articles offer interesting refinements of older data. DeStefano et al (18), recognizing that autism usually did “not have  well demarcated date of onset,”   “compared the distribution of ages at first MMR vaccination” assuming that “if the MMR vaccine increases the risk of autism…then children who are vaccinated at younger ages would have a higher risk of …autism.” (18) This study merely supports the fact that an earlier age compared to a later age of MMR administration does not increase  the risk of autism. While this further negates the Wakefield hypothesis of a short latency, in the larger picture, this study merely supports the probability that a later age of administration does not change the rate of autism.            Much of the energy and detailed analysis has been based on the hypothesis put forward by Wakefield (1), which needed to be addressed. The seminal Oak Brook Report (2) occurred during the first wave of prevalence articles which began to indicate an increased rate of prevalence in AD/ASD was the first major attempt to make sense of the controversy. The Wilson article(4) was merely a collection of the studies focusing on the rejection of  Wakefield’s autism - MMR hypothesis. It is not surprising that in focusing on the measles-autism possibility, other  possibilities were overlooked. PREVALENCE - From Wakefield/measles to other possibilities
            Virtually all studies focused on data analyzed after the introduction of the MMoR in 1971 in the US, but which was changed to MMR27/3 in January, 1979. Single measles vaccines and measles-rubella vaccines were being utilized both inside and outside the US in the 1970’s and into the early and mid-1980’s but with a 50% utilization at best. It appears that no MMR was administered outside the US prior to 1982.
             MMoR      - 1971   -     1/1979  - in US only
             MMR27/3 - 1979   -     present- US
                              - 1982   -    present - Sweden (9)
                              - 1982   -    present - Finland  (13)
                              - 1987   -    present - Denmark (5)
                            - 10/1988 - present - UK

(6) Other countries do not appear to have used either the MMoR version or the RA27/3 strain of rubella (alone or as MR vaccine) at any time prior to the MMR27/3 introduction. This change has not been noted in any of the articles or conferences studying the issue. The prevalence rate needs to be examined from the perspective of a possible MMR27/3-AD/ASD association. Since the MMR27/3 was introduced at different dates in different countries, we will attempt to examine each separately.

SWEDEN

             Sweden is unique in being the only country with all published studies including Gillberg as an author. The MMR27/3 was introduced in Sweden in 1982 (9).  Gillberg documented the prevailing prevalence rate of 4 (‘62 -’76 BrCo) (19); but this later increased to 7.5 by 1984 (20) and 11.6 (1975-’84 BrCo) (10). The latter two studies support the association of AD/ASD with the MMR27/3 introduction. Using an expanded definition of ASD, a ‘74 - ’88 BrCo had a rate of 12 by 1991 and 14 by 1993 (21). An even later study using an ‘88 -’91 BrCo finds the AD + ASD rate of 31 (22); even the smaller AD rate of 10 clearly documents a major change (22). An even higher combined AD+ASD rate of 60 was found for an ‘85 BrCo. (23) FRANCE              Looking at the French experience, the MMR27/3 was introduced sometime during the ‘76-’85 years (24). Fombonne et. al. (25) reviewed a ‘76-‘85 BrCo and found that the AD rate was not significantly different, but the issue of PDD was introduced. The claim was made that the rate of PDD had not increased from the Fombonne and du Mazaubrun ‘92 article, but this earlier article (26) fails to mention PDD. Even more interesting is Figure 1 in which the ‘81 BrCo has a distinct prevalence rate increase between the  ‘85 analysis (26) and the ‘92-93 analysis (25). The latter authors point to a puzzling drop in prevalence rate of AD in the ‘82 - ‘85 BrCo as support of no AD increase post MMR27/3. Three factors weaken this conclusion: 1) the ‘92-‘93 analysis collected data from governmental records, with psychometric data missing in “nearly half” the cases, 2) the retrospective application of  ICD-10 criteria on data collected using ICD-9 criteria, at a time of evolving understanding of PDD in the medical community, and 3) the addition of the PDD category for cases in which “reliable differentiation…was not possible.” Given the confusion of the time, this could have been a reasonable conclusion. But when Fombonne later recognizes the “increased incidence of these (PDD) disorders” (27), then reinterpretation of earlier data is not unreasonable. Viewing the ‘81 cohort in Figure 1 as evidence of an increase in AD clearly evidenced by the shift in prevalence rate supports the claim for an association with the MMR27/3 introduction. The purported drop later could represent a redistribution of cases between the old AD category to the newer PDD category, being sorted out in the late ‘80’s. At the very least, the admitted increase in the PDD incidence must raise some questions.

UK
            The MMR27/3 was introduced in England in October 1988. Webb et. al.(28) demonstrated an increase in AD from 3.3 in a ‘77 -‘79 BrCo to 9.2 in an ‘87-’89 BrCo, coinciding nearly perfectly with this MMR27/3 introduction (assuming children born 10/86 onward being eligible for the vaccine). A similar rate of 8 was found in an ‘88 birth cohort of males by Kaye (11). A mere five years later they demonstrated an increase to 29 in a ‘93 BrCo. The AD rate of 30.8 in a ‘90-’91 BrCo  by Baird et. al. (29) and of 16.8 in a ‘92-95 BrCo by Chakrabarti & Fombonne (27) both confirm a definite increase.
            Of interest, when the PDD rates are added into the latter two studies, 27.1 vs. 45.8, the combined AD + PDD rates become 57.9 (29) and 62.6 (27). The closeness of the combined rates viewed against the widely discrepant  PDD rates support the lack of consensus even among the leading experts of the day regarding categorization.

US
           US studies have been unique in producing lower rates than non-US studies.  Initially, the acceptance and use of the MMR was substantially below 90%. With the change from MMoR (1971 -1978) to MMR27/3 (1979 - present) there is the unique situation of having MMoR supplies being used for two and possibly more years after the introduction of MMR27/3.Therefore, a more accurate rendition or MMoR use would be (1971 - 1978, dwindling through ?1982). Both of these factors could explain the low prevalence rates in earlier studies, including Ritvo et. al. (30) , Burd et. al. (31) and Kirby (32).
           Ritvo compares pre- ‘79 BrCo with an ‘80-’84 BrCo, showing low prevalence rates, even lower than all previous reports. Unfortunately, the ‘80-‘84 BrCo was analyzed when many were less than three years of age, clearly an incomplete analysis of prevalence rates. (This is smilar to the problem with the Madsen study, noted above.) Another study with an ‘85-’87 BrCo has a rate of 4 (32) but is hampered by two factors: 1) reports come from a registry of all developmental disorders with no attempt to either find missed cases or to confirm accuracy and 2) only cases identified by age 4 are counted, which underestimates the prevalence rate. Even more interesting, is the Burd study (31) which found 62% of the AD/ASD cases in the youngest 33% of the population, born (est.) ‘78-’83. The authors found this clearly disproportionate distribution of cases in the younger ages “difficult to explain” which would be true if the MMoR had remained unchanged. The introduction of the MMR27/3 dovetails with the increased percentage of cases in the younger ages.
           Early documentation of an increase in ASD comes from Dales (12) showing an increase from 4 in ‘80 to 20.8 in ‘94, and from Bertrand (33) who demonstrated even higher rate of 40 for an ‘88-’95 BrCo, but did not break the rate down by years. Croen et al (34), in the first study using successive BrCo‘s, demonstrated an increased rate of AD alone from 5.8 in 1987 to 14.9 in 1994. Croen tries to explain the increase away as merely a reclassification of MR cases, but the graph supports this possibility only  from ‘87 to ‘90, while clearly contradicting this thesis from ‘90 to ‘94. While a few cases may have been misclassified earlier, this hypothesis has not been widely accepted or substantiated elsewhere.
          Yeargin-Allsopp et al (35) substantiate the higher ASD rates of 47 for 8 year olds, 42 for 5 year olds, 31 for 4 year olds and 19 for 3 year olds, supporting the delayed documentation of ASD in younger cohorts (See comment above regarding Ritvo and Madsen studies) as well as higher prevalence rates at least after ‘86 birth cohorts. Recent studies add value by demonstrating ASD in individual BrCo’s followed over distinctly separate years, not just in one study year or a cohort spanning several years. Gurney et al.(36)  sheds light in their Figure 3 which shows a virtually flat line with no increased ASD in the 1981-1982 birth cohorts, corresponding to the MMoR time frame with dwindling use after 1979. All birth cohorts from 1983-1984 through 1993-1994 show progressively dramatic increases in ASD prevalence rates. Later, Barbaesi et al. (37) shows an 8.2 fold increase in the incidence of “research-identified autism” from 5.5/100,000 in 1980 -1983 to 44.9/100,000 in 1995 -1997. In addition,  though slightly more confusing, their graphs in Figure 3 demonstrate a slightly mixed picture of a borderline increased  prevalence with 1979-1984 BrCo and 1982-1987 BrCo but clearly increased rates for 1985-1990 and later BrCo’s - again supporting the time frame of dwindling MMoR use after 1979.  Even in the US, studies show definite increases in the prevalence of ASD after 1985, with mixed results coming from the early 1980’s.

JAPAN
          A recent study by Honda et. al. demonstrated that the withdrawal of the MMR27/3 could not “explain the rise over time in the incidence of ASD” which continued even after no MMR27/3 was no longer used. (38)Not clarified in the article was exactly what was used in place of the MMR27/3. It was “parents’ choice” as to which of the individual monovalent vaccines were used. “Most of the children who were not given the MMR were given monovalent measles and monovalent rubella vaccine separately, with an interal of at least four weeks.” (Personal communication) (39) The study established that separating the individual components of the MMR27/3 would not result in a decreased incidence in autism. DISCUSSION Changed rates of prevalence
          The initial focus on the measles portion of the MMR, without recognizing the difference between the MMoR and MMR27/3, produced some interesting and some early weak studies, not surprising given a new hypothesis. The studies focusing on the lack of temporal relationship were either excessively limited in follow-up (5,13,14, 30), too focused on accepting only a clearly defined peak or “significant clustering” (17) or merely rejecting the theory that later administration of the MMR was not protective (18). The studies which focused on the lack of a parallel rise in MMR and autism (11,12) fail to address the issues of
              a) catch up doses being given to children which were older than the routine 12-15 months of age for MMR administration but whom would have received the “new” vaccine, at least up to 2 years of age and possibly older or b) delayed penetration of the MMR27/3 as occurred in the US. Though worthwhile data in the measles - MMR debate, they fail to be rigorous enough to fully reject the MMR27/3 - autism hypothesis.
           Any discussion of prevalence rates is complicated by changing criteria from Kanner (1943), to Rutter (1978), to DSM III (1980),  DSM-R III (1987) and DSM IV (1994), but within these criteria the persistent use of repetitive stereotyped behavior and impaired reciprocal social interaction has been very stable.  With this evolution of diagnostic criteria has come the appreciation of differentiating core AD from a  more varied picture, termed either ASD or PDD. In addition, methods of establishing rates of prevalence were  complicated by varied methods of data collection and the concern regarding the completeness of the data. Coincidentally with this changing picture of autism, the rate of autism appeared to be increasing sometime during the 1980’s. Early debates attempting to make sense of the data were stymied by inconsistent rates coming in from around the world, as well as a lack of consistent methodology.
           Changing criteria and clinical sensitivity undoubtedly have clouded the picture, at least from the early 1980’s before which it is doubtful that any true increase occurred. Wing’s review (40) in 1993 recognizes the variation in prevalence rates, and the paucity of studies addressing the significance of years of birth at that time. Wing excluded a varied awareness of autism, age of onset criterion, and methods of examination as possible explanations for the differences in prevalence. A later review by Gillberg & Wing (41) in 1999 recognized the “higher prevalence figures reported since about 1985” with the exception of two US studies with “atypically low rates.“ They found “no convincing evidence” that this increased rate was due to “any …particular diagnostic system.” They concluded that autism could “no longer be conceptualized as an extremely rare disorder.”  Gillberg, in the Kadesjo et. al. article (23) in 1999, uses virtually identical language - “not extremely rare” while clinging to the belief that “there is remaining doubt as to whether the prevalence…has increased or not.“ (23) The proposed solution was “larger scale cross-cultural population studies.“ (23)
          A thorough and detailed review of the epidemiology of autism by Fombonne (24) in 1999 delineates the difficulties of comparing data. The median prevalence rate had increased from 4.3 during 1966-1988 to 7.2 during 1989 - 1998, attributing the increase to “an improved recognition…together with a broadening of the diagnostic concept and definitions“. (p. 777)  Indeed, while initially denying that this could “be attributed to an increase in the incidence” of AD, in 2001 Fombonne reversed this opinion after documenting an increased prevalence rate of 16.8 for an English 1992 - 95 BrCo  (27) but hedged the conclusion, citing that “vigorous testing” would be needed to prove an increased incidence (as opposed to prevalence). Even as they reject the hypothesis of an increased rate of regression with MMR, in an early study spanning 41 years, one is struck by a dramatically increased number of cases from 99 over 25 years (1954-79 BrCo) to 97 over a mere 4 years (1992-1995BrCo). (8)
             Not until the late 1990’s, has there been a broader consensus of understanding and knowledge of the increased prevalence of autism. As rates of ASD were reported in the 40 or greater range by Bertrand (33) in 2001 and later Yeargin-Alsopp (35) in 2003, it was becoming obvious that something had changed to increase the prevalence of  ASD. Whether the increased rates reflected an increase in core autism or an increase in the PDD sector, or both, something had appeared which was not previously present.

Timing of change
          Without factoring in the introduction of the MMR27/3, it reasonable to see no consistent pattern. If the introduction of the MMR27/3 vaccine was a factor in the increase prevalence /incidence of AD/ASD, then it is important to recognize that the vaccine was introduced in different countries at different dates. After factoring in the timing of the introduction, then some consistencies seem to emerge with few exceptions. Each country must be viewed individually.           Sweden’s rate goes from 4.0 pre MMR27/3 vaccine 1982 introduction, to 7.5 (’75-’84 BrCo - mostly pre MMR27/3) to 10 for AD and 31 to 60 for ASD. (See Sweden Table.) In the UK, the rate increases from less than 5 in the 1970’s and early 1980’s to 8 in a 1988 BrCo (1st year of MMR27/3 introduction) to 29 (‘93 BrCo)(11) and a similar rate of 30.8 (1990-1991 BrCo) by Baird (29). In the US, the 8.2 fold increase in “research identified” AD, going from 5.5 (‘80-’83 BrCo) to 44.9 (‘95-97 BrCo)/100,000 substantiates a similar increase.  (Why  the rate is per 100,000 vs. the usual 10,000 is unclear but the rate increase is not unclear.)
         A review of the literature indicates that while many cases can be diagnosed by three years of age, the reality is somewhat different. Stopping an analysis too soon after 18 months, when most MMR27/3 doses would have been given, will lack sensitivity. The additional factor of Asperger’s syndrome, whether included or not, complicates the picture but on the whole most agree that the addition of Asperger’s would only increase the numbers significantly.
        Moreover, the reality of catch up immunization occurring in BrCo’s born two to three years before the MMR27/3 introduction would tend to “contaminate” numbers giving the impression of an increased incidence occurring in children born before its introduction, thus explaining the low grade increased prevalence seen in Taylor (Figure 1)(6) preceding the MMR27/3 introduction in 1988. The slow increase seen in Dale’s chart (Figure, p. 1184) (12) from the US could be explained by both the persistence of MMoR supplies for some time after the introduction of the MMR27/3. The slow increase in prevalence in both studies could also well reflect the slow recognition of the increased incidence of autism, especially the PDD spectrum. While the numbers are not crystal clear and full appreciation of AD/ASD was slow in establishing itself, the trend fits.
                    Conclusion
The increased rate in AD/PPD prevalence first appearing in the mid 1980’s and becoming more established and recognized in the 1990’s is definite. Rejecting the MMR-autism association is based on at least two false assumptions:
               1) The MMR remained unchanged during the period of analysis.
               2) No catch up immunizations were given to older children after the introduction of the MMR in countries outside the US. The data has not recognized the introduction of the MMR27/3 in the US after eight years use of the older MMoR. Neither has the literature addressed the fact that the MMR27/3 was introduced de-novo in non-US countries,  confounding any data comparison. All non-US countries have used only the MMR27/3 and this was introduced at different dates in the 1980’s. When the available data is separated into pre-MMR27/3, transitional MR27/3 and post MMR27/3 categories, (see charts) the increased prevalence in AD/PDD is pretty clear. The bulk of data points to a positive association  of AD/ASD with the introduction of the MMR27/3 vaccine, most specifically the Rubella27/3 portion. It is time for this factor to be addressed.

References:
1. Wakefield AJ, Murch SH, Anthony A, et al. Ileal-lymphoid-nodular hyperplasic, non-specific colitis, and
pervasive developmental disorder in children. Lancet. 1998; 351:637-641.
2. Halsey, N, Hyman, S and the Conference Writing Panel. MMR Vaccine and Autistic Spectrum Disorder: Report
from the New Challenges in Childhood Immunizations Conference convened in Oak Brook, Illinois, June 12-
13,2000. Pediatrics 107: p. e84
3. http://www.cdc.gov/epo/mmwr/preview/mmwrhtml/00053391.htm
4. Wilson, Kumanan et al. Association of Autistic Spectrum Disorder and the MMR Vaccine. Arch Pediatr Adolesc
Med . 2003: 157:628-634.
5. Madsen, Kreesten Melgaard et al. A Population-Based study of  MMR vaccination and autism. NEJM. 2002:
1477-1482.
6. Taylor, Brent et al. Autism and MMR vaccine: no epidemiological evidence for a causal association. The Lancet.
1999; 353: 2026-2029.
7. Taylor, Brent et al. MMR vaccination and bowel problems or developmental regression in children with autism:
population study. BMJ. 2002; 393-396.
8. Fombonne, Eric and Chakrabarti, Suniti. No Evidence for a New Variant of MMR-Induced Autism. Pediatrics.
2001: 108: 58- .
9. Gillberg C and Heijbel H. Commentaries: MMR and autism. Autism. 1998; 2(4):423-424.
10. Gillberg, C., Stefenburg, S., and Schaumann, H. Is Autism More Common Now Than Ten YearsAgo? British
Journal of Psychiatry. 1991; 158: 403-409.
11. Kaye, James K. et al. MMR vaccine and the incidence of autism recorded by general practitioners: a time trend
analysis. BMJ. 2001; 322:460-3.
12. Dales, L, Hammer, SJ & Smith, N. Time Trends in Autism and in MMR Immunization Coverage in California.
JAMA. 2001; 285: 1183-85.
13. Patja, Annamari et al. Serious adverse events after MMR vaccination during a fourteen-year prospective follow-
up. Pediatr Infect Dis J. 2000; 1127-34.
14. DeWilde, S. et al. Do children who become autistic consult more often after MMR vaccination? Br J Gen Pract.
2001; 51: 226-227.
15. Farrington, C. Paddy et al. MMR and autism: further evidence against a causal association. Vaccine. 2001; 19:
3632-3635.
16. Roger, JH. The MMR question. Lancet.2000; 356:160-1 corresp.
17. Makela, Annamari et al. Neurologic disorders after MMR vaccination. Pediatrics. 2002; 110:957-63.
18. DeStefano, Frank et al. Age at First MMR Vaccination in Children with Autism and School-Matched Control
Subjects: A Population-Based Study in Metropolitan Atlanta. Pediatrics.2004; 113: 259-266.
19. Gillberg,Christopher. Infantile autism and other childhood psychoses in a Swedish urban region.
Epidemiological aspects. J. Child Psychiat. 1984; 25, 35-43.
20. Steffenburg, Suzanne and Gillenberg, Christopher. Autism and Autistic-like Conditions in Swedish Rural and
Urban Areas: A Population Study. British Journal of Psychiatry.  1986; 149: 81-87.
21. Nordin, V., and Gillberg, C. Autism Spectrum Disorders In Children With Physical Or Mental Disability Or
Both. I: Clinical and Epidemiological Aspects. Developmental Medicine and Child Neurology. 1996; 38, 297-313.
22. Arvidsson, T., Danielsson, B., Forsberg, P., Gillberg, C., Johansson, M., Kjellgren, G. Autism in 3-6-year-old
children in a suburb of Goteborg, Sweden. Autism. 1997; 1, 163-173.
23. Kadesjo B, Gillberg C, and Hagberg B. Brief Report: Autism and Asperger Syndrome in Seven-Year-Old
Children: A Total Population Study. J. Autism and Developmental disorders. 1999; 29:327-331.
24. Fombonne, Eric. The epidemiology of autism. Psychological Medicine. 1999; 29,769-786.
25. Fombonne, Eric, Du Mazaubrun, Christiane, Cans, Christine, and Grandjean, Hélène. Autism and Associated 
Medical Disorders in a French Epidemiological Survey. J. Am.Acad.Child Adolesc. Psychiatry. 1997; 36(11): 1561-
1569.
26. Fombonne E and du Mazaubrun Prevalence of infantile autism in four French regions. C. Soc Psychiatry
Psychiatr Epidemiol. 1992; 27: 203-210.
27. Chakrabarti S and Fombonne E. Pervasive Developmental Disorders in Preschool Children. JAMA. 2001: 285:
3093-3099.
28. Webb EVJ, Lobo S, Hervas A, Scourfield J, Fraser WI. The changing prevalence of autistic dosorder in a Welsh
health district. Developmental Medicine & Child Neurology. 1997; 39: 150-152.
29. Baird G, Charman T, Baron-Cohen S, Cox A, Swettenham J, Wheelwright S, and Drew A. A Screening
Instrument for Autism at 18 Months of Age: A 6-Year Follow-up Study. J. Am. Acad. Child Adolesc. Psychiatry.
2000; 39(6): 694-702.
30. Ritvo ER, Freeman BJ, Pingree C, Mason-Brothers A, Jorde L, Jenson WR, McMahon WM, Petersen B, Mo A
and Ritvo A. The UCLA-University of Utah Epidemiologic Survey of Autism: Prevalence. Am J Psychiatry. 1989;
146: 194-199.
31. Burd, L, Fisher, W and Kerbeshian,J. A Prevalence Study of Pervasive Developmental Disorders in North
Dakota. Acad. Child Adol. Psychiat. 1987; 26,5:700-703.
32. Kirby RS, Brewster MA, Canino C and Pavin M. Early Childhood Surveillance of Developmental Disorders by
a Birth Defects Surveillance System: Methods, Prevalence Comparisons, and Mortality Patterns. J Dev Behav
Pediatr. 1995; 16: 18-326.
33. Bertrand J, Mars A, Boyle C, Bove F, Yeargin-Allsopp M, and Decoufle P. Prevalence of Autism in a United
States Poulation: The Brick Township, New Jersey, Investigaton. Pediatrics. 2001; 108: 1155-1161.
34. Croen LA, Grether JK, Hoogstrate J and Selvin S. Journal of Autism and Developmental Disorders. 2002;
32:207-215.
35. Yeargin-Allsopp M, Rice C, Karapurkar T, Doernberg N, Boyle C and Murphy C. JAMA. 2003: 289: 49-55.
36. Gurney JG, Fritz MS, Ness KK, Sievers P, Newschaffer CJ and Shapiro EG. Analysis of Prevalence Trends of
Autism Spectrum Disorder in Minnesota. Arch Pediatr Adolesc Med. 2003; 157: 62-627.
37. Barbaresi WJ, Katusic SK, Colligan RC, Weaver AL and Jacobsen SJ. The Incidence of Autism in Olmsted
County, Minnesota, 1976-1997. Arch Pediatr Adolesc Med. 2005; 159: 37-44.
38. Honda, H, Shimizu, Y, and Rutter, M. No effect of MMR withdrawal on the incidence of autism: a total
population study. Journal of Child Psychology and Psychiatry. 2005; 46(6): 572-579.
39. Honda, Y. Personal communication. 9/15/2006.
40. Wing , Lorna. The Definition and Prevalence of Autism: A Review. European Child and Adolescent Psychiatry.
1993; 2; 61-74.
41. Gillberg, C and Wing, L. Autism: not an extremely rare disorder. Act Psychiatr Scand . 1999; 399-406.