News and Commentary

Molecular Psychiatry (2003) 8, 133–134. doi:10.1038/

MMR and autistic enterocolitis: consistent epidemiological failure to find an association

E Fombonne1 and E H Cook Jr2

  1. 1Department of Psychiatry, Montreal Children's Hospital, Montreal H3Z 1P2, Canada
  2. 2University of Chicago, Chicago, IL 60637, USA

Correspondence: Dr E Fombonne, Department of Psychiatry, Montreal Children's Hospital, 4108 Ste-Catherine West, Montreal, Canada H3Z 1P2. E-mail:

In 1998, the authors of a study of 12 children with pervasive developmental disorders (PDD) referred to a gastroenterology unit in a London (UK) hospital claimed that a new variant of autism, induced by immunization with the combined measles, mumps and rubella vaccine, had been identified1 and was responsible for increasing rates of PDD worldwide.2 Subsequent studies from the same group of investigators3,4 on a larger sample claimed to have identified biological evidence of the persistence of measles virus in peripheral blood white cells and gut biopsies. These studies were criticized on the basis of a lack of clear definition of the clinical phenotype in the study subjects5 and of many laboratory procedural problems, which called for independent replication of the findings.6

Several epidemiological investigations have tested the association between trends in rates of autism and either the introduction of measles–mumps–rubella (MMR) vaccine7 or variations in MMR uptake,8,9 and they all failed to demonstrate an association. Further analyses have examined whether or not regression in the course of autism10,11 or health-seeking behavior by parents12 clustered after MMR vaccination, and no evidence of time clustering was found. More recently, a large population-based cohort study of over 500,000 Danish children has shown no increase in the incidence of autism in children exposed to MMR compared to unexposed children.13 However, it could be argued that, since these studies were taking any form of autism as their case definition, an association applying specifically to a small subset of autism cases might have gone undetected. Although the clinical phenotype in question varied from account to account, Wakefield et al3 had argued once that the most distinctive features of their new autism variant consisted of regression in the developmental course associated with gastrointestinal symptoms, although precise delineation of this 'autistic enterocolitis' has been largely missing.5,11

In this context, the recent publication by Taylor et al14 adds new empirical evidence in the debate. Taking Wakefield's hypotheses strictly, they investigated in a long time series of autism cases whether or not the regressive form of autism and the incidence of gastrointestinal symptoms had increased over time, looking particularly at a step-up in rates of these specific syndromes after the 1988 mass introduction of MMR in the UK. In this large (n=473) sample, 17% of children were reported to have suffered from gastrointestinal symptoms lasting 3 months or more, and 25% were reported to have lost skills or regressed in their development, findings that are consistent with other epidemiological surveys.11 No change over time was reported for both regression and occurrence of gastrointestinal symptoms, and results were similar for both typical and atypical forms of autism. Furthermore, the rates of both problems did not differ in children for whom parents first became concerned before the MMR immunization, as compared to those for whom parents became concerned after the MMR immunization. The latter group should have exhibited an increased frequency of regression and/or bowel problems had there been an association with MMR. Taylor et al found a significant association between regression and bowel symptoms, but only 6.6% of the entire sample showed this combined set of problems and no association with exposure to MMR or increase over time was detected.

Although the authors did not raise this possibility, the fact that bowel symptoms were more often reported in autistic children with regression might reflect, in a retrospective analysis of medical records, the tendency for autistic children who exhibit regression to be more often subjected to in-depth medical investigations and surveillance, thereby increasing the likelihood of a spurious association. It is noteworthy that in another epidemiological series where children were prospectively identified, no association was reported between bowel symptoms and regression according to both contemporaneous parental and medical reports.11 Similarly, no association between autism and gastrointestinal disorders was found in a nested case–control study where gastrointestinal symptoms and disorders were prospectively identified before the diagnosis, although information about regression was not available in this investigation.15

The strengths of the Taylor et al investigation lie in a large sample, with a systematic mode of ascertainment, encompassing a 20-year interval, allowing adequate control for confounding factors and for testing effects according to subtypes of autism. In that study, immunization dates were obtained from computerized vaccination records and information bias could therefore not occur. This study provides further evidence that the phenotype postulated by Wakefield has a low base rate and does not account for more than a small proportion of autism cases. An immediate conclusion is that this so-called variant of autism cannot explain the admittedly impressive rise in autism rates. Furthermore, this phenotype is not associated with MMR exposure in two large-scale epidemiological studies of representative samples of children with autism or PDD.

All studies have thus far failed to support the claims of the Royal Free Hospital group. The initial observations were made on unrepresentative samples of children (self-) referred to a pediatric gastroenterology department, a situation well known to lead to potentially serious biases in case–control hospital-based studies. Interestingly, it should be recalled that the same author made claims about an association between measles viruses and inflammatory bowel disorders in the early 1990s, which led similarly to huge public and professional concerns. Again, follow-up studies by independent investigators failed to support the initial claims. How the initial studies of Crohn's disease and ulcerative colitis relate (or not) to the most recent 'autism enterocolitis' claims has been, surprisingly, left obscure by Wakefield. At any rate, there appears to be no association between inflammatory disorders and autism.16

How many more well-powered epidemiological investigations of representative samples will be necessary for this hypothesis to be completely discarded? Obviously, the accumulation of studies that fail to reject the null hypothesis of no association between MMR and autism is not appraised in the same way as a single 'positive' but intriguing finding deriving from a small clinical sample of unknown representativeness. Far from applying only to the lay audience, these biases appear to affect scientists and journal editors as well. As Jonathan Swift said: 'Falsehood flies and the truth comes limping after; so when men come to be undeceived, it is too late: the jest is over and the tale has had its effect'.17 There are other recent examples of poor science leading to long-lasting deleterious consequences on public health and social policies; for example, the breast implant controversy in the early 1990s provides an interesting parallel where scientific evidence failed to influence popular beliefs and legal actions once these were set earlier in motion albeit on much weaker evidence.18 In the case of MMR and autism, early failures in the peer review process can be identified as contributing factors to that story. Thus, the initial claims that onset of autism symptoms was observed within days after immunization were accepted at face value, whereas these were all based on retrospective and unvalidated accounts and experts in the field know very well that dating the onset of symptoms in days is not possible. Regression in autism was presented as a new phenomenon whereas it had been known to affect a quarter to a third of children with autism for decades. Later, in an attempt to adjust to the comments made by scientific peers in various committees, the hypotheses were changed (ie, the regression could now occur years after the immunization...), although confusion persisted in many aspects of the research (ie, children with dyslexia, ADHD and schizophrenia were included in Wakefield's samples3 and no precise clinical description was ever given of the putative 'autistic enterocolitis' phenotype5). It is now hoped that more attention will be given by reviewers, editors and researchers to basic methodological flaws in the articles that allowed the MMR-autism controversy to expand and to what must now be regarded as a consistent failure to substantiate the original claims in robust epidemiological studies.



  1. Wakefield A et al. Lancet 1998; 351: 637–641. | Article | PubMed | ISI | ChemPort |
  2. Wakefield A. Lancet 1999; 354: 949–950.
  3. Wakefield A et al. Am J Gastroenterol 2000; 95: 2285–2295. | PubMed |
  4. Uhlmann V et al. J Clin Pathol: Mol Pathol 2002; 55: 0–6.
  5. Fombonne E. Mol Pathol 2002 (letter, 27 May).
  6. Halsey N. Mol Pathol 2002 (E-letter 15 March).
  7. Taylor B et al. Lancet 1999; 353: 2026–2029. | Article | PubMed | ISI | ChemPort |
  8. Dales L et al. J Am Med Assoc 2001; 285: 1183–1185. | Article | ISI | ChemPort |
  9. Kaye J et al. Br Med J 2001; 322: 0–2.
  10. Farrington C et al. Vaccine 2001; 19: 3632–3635.
  11. Fombonne E, Chakrabarti S. Pediatrics 2001; 108: E58. | Article | PubMed | ChemPort |
  12. DeWilde S et al. Br J Gen Pract 2001; 51: 226–227. | PubMed | ISI | ChemPort |
  13. Madsen K et al. N Engl J Med 2002; 347: 1477–1482. | Article | PubMed | ISI |
  14. Taylor B et al. Br Med J 2002; 324: 393–396. | Article |
  15. Black C et al. Br Med J 2002; 325: 419–421.
  16. Fombonne E. Lancet 1998; 351: 955.
  17. Swift J. The Examiner, 1710; 15: 9 November.
  18. Angell M. Science on Trial. WW Norton & Company: New York, 1996.