Introduction

Prior to the introduction of vaccines in Canada, measles, mumps, and rubella were associated with substantial morbidity and premature death in children. Since the implementation of universal infant measles-mumps-rubella (MMR) vaccination programs, Canada has seen a marked reduction in the incidence of these diseases.[1] However, several published reports have questioned the safety of MMR vaccination. Researchers mainly affiliated with the Inflammatory Bowel Disease Study Group (IBDSG) in the United Kingdom suggest that MMR vaccination is associated with chronic inflammatory bowel disease (IBD) which, they further allege, can lead to autism. This paper reviews and summarizes the evidence of whether MMR vaccination is causally associated with IBD and subsequent development of autism.

Methods

Between November 2000 and February 2001, we conducted an Internet Grateful Med search of Medline for publications from 1980 to December 2000 related to MMR vaccination or MMR infection and autism. Publications were identified by a combination of key words including: autism, autistic disorder, measles, measles virus, measles vaccination, MMR, measles-mumps-rubella vaccine, immunization, and vaccination. We included articles if the title or abstract referred to both MMR infection, MMR vaccination, or one of its component parts, and autism. Concurrently, we conducted a similar literature search for published articles from 1996 to December 2000 that examined the association between MMR vaccination or MMR infection and IBD. The combinations of key words included Crohn’s disease, ulcerative colitis, inflammatory bowel disease, measles, measles infection, vaccination, immunization, perinatal, in utero infection, MMR, and measles-mumps-rubella vaccine. We included articles published since a 1997 review of measles infection and IBD[2] and, if the title or abstract referred to both MMR infection, MMR vaccination or one of its component parts, and IBD. For both literature searches, secondary references were also obtained using similar selection criteria. Conference abstracts or proceedings and unpublished reports were not reviewed. The quality of evidence was evaluated based on study design, use of appropriate controls, sample size, data collection methods, control for confounding factors (i.e., age, sex, socioeconomic status, other chronic conditions or illness, past medical history) and appropriate analysis.

MMR vaccination and autism 

In 1998, a study by Wakefield and colleagues of the IBDSG, Royal Free Hospital, London, England, alleged a causal link between MMR vaccination and autism. From a case series, Wakefield and colleagues describe the investigation of 12 consecutively referred children with gastrointestinal symptoms who were also diagnosed with autism.[3] Parents or a physician had temporally linked the onset of behavioral symptoms to MMR vaccination in eight of 12 children, although four of the 12 children had behavioral problems before gastrointestinal symptom onset. The authors speculate that ileal-lymphoid-nodular hyperplasia (ILNH), a non-specific colitis, causes malabsorption of vitamins and nutrients, or an increase in gut permeability to protein, in either case leading to the development of autism. The study had several major methodologic shortcomings including: a sample size too small to assess statistical significance, selection bias from use of a highly select patient population that was not generalizable, and absence of a control group. The study was highly susceptible to the recall bias of parental reporting, and the investigators were unable to provide corroborative microbiologic evidence to support their hypothesis. This study provides little evidence to support a hypothesis that MMR vaccination causes autism.

To address shortcomings of this case series, the IBDSG subsequently included 48 additional cases along with the original 12 and also included 42 controls, who were referrals to their clinic, in whom IBD had been ruled out.[4] This study suffers many of the same methodologic limitations of the investigators’ previous study, including selection bias in cases and in controls, failure to control for confounding variables, and failure to isolate measles virus from bowel tissue. In addition, cases and controls differed in their baseline characteristics and the study did not report the results of any statistical tests of association.

However, several population-based studies provide evidence that MMR vaccination is not associated with autism. Re-analysis of a population-based study by Gillberg and colleagues found no significant difference in autism after introducing MMR vaccine in Sweden.[5] A population-based study by Taylor and colleagues investigated the MMR vaccination status of 498 autistic children in England. This study employed sound epidemiologic methods, particularly with respect to adequate sample size, thorough and consistent ascertainment of cases of autism and prior MMR immunization status, and appropriate analysis.[6] An increase in cases of autism was noted by year of birth from 1979 to 1992; however, no incremental increase in cases was observed after the introduction of MMR vaccination. There was no difference in age of autism diagnosis and MMR vaccination status (relative incidence (RI) 0.94 [95% confidence interval (CI) = 0.60– 1.47]) and there was no temporal clustering between onset of autism and MMR vaccination within the following 1 to 2 years (RI 1.09 [95% CI = 0.76–1.52]). Furthermore, from a general practice research database, Kaye and colleagues identified 305 children diagnosed with autism who were ≤12 years of age. A time-trend analysis found no correlation between prevalence of MMR vaccination and the incidence of autism in each birth cohort from 1988 to 1993.[7] Patja and colleagues reviewed surveillance reports of MMR vaccine-associated adverse events since 1982 in Finland. There were no reports of inflammatory bowel disease or autism over the study period, during which more than 3 million doses of MMR were administered. However, vaccine adverse event reporting identify acute events more easily,[8] and the likelihood of autism or IBD being reported as a vaccine adverse event is in all probability quite small. Finally, a review by Fombonne of published epidemiological surveys of autism between 1966 and 1998 found no reports of an association with inflammatory bowel disease, including Crohn’s disease, or with measles or mumps infections.[9]

Measles virus and inflammatory bowel disease

IBD is a hypothesized intermediary condition for the association between MMR vaccination and autism. Studies have reported conflicting results concerning the detection of measles virus in the intestinal tissue of individuals with IBD. Initial clinical studies[10-13] reported the presence of measles virus in the intestinal tissues of IBD cases. However, this finding has not been corroborated using sensitive reverse transcriptase polymerase chain reaction (RT-PCR), which, thus far, has failed to detect measles virus in any intestinal tissues of IBD cases.[14-18]

Other studies have evaluated individuals for serologic evidence of persistent circulating measles antibody. In a case-control study by the IBDSG, a significant increase in increased serum measles IgM levels in individuals with Crohn’s disease was reported.[19] Again, this finding has not been corroborated by other laboratories, where serologic evidence of increased levels of measles-specific immunoglobulins in affected Crohn’s disease cases has not been found.[20-23] Iisuka and colleagues also reported that detection of a monoclonal antibody to measles-related antigen in bowel tissue was not unique to Crohn’s disease, although baseline characteristics differed between cases and controls.[24] Kawashima and colleagues studied a cohort of nine children drawn from cases diagnosed by Wakefield and colleagues, with ILNH and reported measles virus detection (H gene region) by RT-PCR in peripheral blood mononuclear cells from three of nine cases with ILNH, one of eight cases with Crohn’s disease, and one of three cases with ulcerative colitis, compared to none of the controls.[25] The methodologic weaknesses plaguing the earlier Wakefield study, from which these subjects were drawn, also compromise the validity of this study and its results. This finding could not be confirmed by Afzal and colleagues, who targeted three gene regions of the measles virus genome (H, M, and N) by RT-PCR, using inflamed and normal bowel tissue of Crohn’s disease patients.[26]

Measles vaccination and inflammatory bowel disease

Feeney and colleagues studied 140 cases with IBD and 240 controls.[27] Vaccination history was ascertained from physician and community health records. No association was found between measles vaccination and IBD, with a reported odds ratio (OR) of 0.97 (95% CI = 0.64–1.47). A retrospective analysis of 7616 members of a British cohort study by Morris and colleagues found no significant association between monovalent measles vaccination status and IBD by 26 years of age after controlling for confounding factors.[28] In fact, the point estimates were more suggestive of a protective effect, with adjusted odds ratios for Crohn’s disease and ulcerative colitis of 0.67 (95% CI = 0.27– 1.63) and 0.57 (95% CI = 0.20–1.61), respectively.

Measles infection and inflammatory bowel disease

Studies have examined the association between perinatal measles infection and IBD. Since perinatal measles infection is quite rare, many of these studies are ecologic in design or comprise small numbers of cases identified from records of a single hospital. Two studies by Ekbom and colleagues found an association between perinatal measles infection and subsequent IBD. The first study, which was ecologic in design, compared the observed to expected number of IBD cases born after measles epidemics from 1945 to 1954, reporting a standardized incidence ratio of 1.46 (95% CI = 1.11–1.89).[29] This study did not control for confounding factors and included probable cases of IBD. The second study, a case series, identified measles infection in four of 25 000 deliveries at a hospital and found four children with Crohn’s disease. The measles virus antigen was detectable in intestinal tissue using electron microscopy in three of the four children.[30] The study did not compare this finding with the frequency of detection of measles virus antigen in children not diagnosed with Crohn’s disease. Subsequent studies have found no association between perinatal measles infection and IBD. A chart review of all pregnancies between 1935 and 1985 by Pardi and colleagues identified seven cases of perinatal measles infection out of 67,912 live births, with no evidence of IBD after a mean follow-up of 38 years.[31] Using hospital records, Nielsen and colleagues identified 33 women with in utero exposure to measles virus. Among the 26 offspring identified, no association between in utero exposure to measles virus and Crohn’s disease was found, although outcome ascertainment was dependent upon the sensitivity of case definition of IBD and completeness of the hospital discharge register.[32] An ecologic study by Haslam and colleagues found no association between birth during measles epidemics between 1951 and 1967 and increased incidence of Crohn’s disease.[ [9]33]

Investigators have also looked at measles infection in infancy and childhood and subsequent development of inflammatory bowel disease. In an ecologic study comparing observed to expected rates of IBD in Olmsted County, Minnesota, Pardi and colleagues noted a 3-fold higher incidence of Crohn’s disease and ulcerative colitis among children who had reported measles infection before 5 years of age.[34] However, with only a 57% participation rate, this study was vulnerable to volunteer and recall bias. In addition, the measles case definition used was based on clinical symptoms as opposed to laboratory confirmation. Montgomery and colleagues conducted a retrospective cohort analysis of self-reported IBD in persons 26 years of age. Just over 7000 of the original cohort of 16 000 had a subsequent physician-confirmed diagnosis of IBD. They then examined the association of IBD in five documented childhood infections. A significant association was noted between the occurrence of measles and mumps infections within the same year and subsequent ulcerative colitis (OR 7.47, 95% CI = 2.42–23.06) and Crohn’s disease (OR 4.27, 95% CI = 1.24–14.46) independent of sex, social class at birth, household crowding in childhood, and family history of IBD.[35] Notably, all but one of the study participants were diagnosed with IBD after 16 years of age, presumably many years after having been affected with measles and mumps within the same year. There was no significant association between measles infection at a younger age and later IBD, and no association between monovalent measles vaccination and IBD. However, the choice of insulin-dependent diabetes mellitus individuals as controls may have biased the outcome, since vaccination may be protective in this population under some circumstances.[36] In addition, a recent nested case-control study, matched on gender and social class, found no association between IBD and measles infection, independent of potential confounding factors (OR for Crohn’s disease 1.09, 95% CI = 0.36–3.52).[37]

Discussion

Review of the literature since the publication of Wakefield’s paper in 1998, in which MMR vaccination was alleged to cause autism,[3] reveals little evidence to support this hypothesis. A recent review by the Medical Research Council of the United Kingdom also found insufficient evidence to support a link between MMR and autism.[38] Nor was any new evidence presented in scientific testimony at a hearing of the United States Committee on Government Reform that examined this issue in 2000.[39] The hypothesis that MMR causes autism is based primarily on studies undertaken by a single group of researchers—the IBDSG—whose studies have important epidemiologic weaknesses and whose findings have, for the most part, not been confirmed by other investigators. In fact, studies that have looked specifically at the association between MMR vaccination and autism have generally found either no evidence of an association, or evidence supporting a non-association. Similarly, there is insufficient evidence to support a link between MMR vaccination and IBD and most significantly, measles virus has not been isolated in the tissue of patients with IBD by RT-PCR.

The etiology of autism and IBD is unknown, although evidence points to both diseases having multiple potentially interactive causes, including genetic factors.[9,40,41] To properly evaluate whether MMR vaccination causes autism, studies are required that adequately control for potential confounding factors, such as age, sex, family history, dietary factors, other infectious agents, and socioeconomic factors. Referral bias should be minimized by random, population-based recruitment of cases and controls. Ascertainment of cases and controls requires consistent application of a valid, reliable case definition of autism, and evidence of prior MMR vaccination should be based on documented evidence and not reliant on parental recall. With high population rates of MMR immunization, it may be difficult to recruit sufficient numbers of unimmunized subjects into such studies. An alternative approach, if a dose response relationship exists between MMR vaccination and autism, is to compare health status after one- versus two-dose MMR immunization.

Conclusion

The evidence does not support a causal association between MMR vaccination and autism. Although there may be biologic plausibility for an association, there is a lack of evidence in five of the classic attributes of causality: consistency, strength of the association, specificity, dose response, and experimental evidence. Whereas the risks of disease and complications of disease in the unvaccinated are known and real, the role of MMR vaccination as an initiator or promoter of autism remains unsubstantiated. The results of this review support the remarkable safety record of MMR vaccine. This, along with the demonstrable effectiveness of MMR vaccine in virtually eliminating these diseases in Canada, provide powerful evidence in support of universal infant MMR vaccination programs.

Traditional surveillance of vaccine safety does not readily identify possible associations between immunization and chronic disease. The authors feel this issue illustrates the need for a timely, ongoing coordinated global approach to evaluating complex issues related to vaccine safety.

Acknowledgments

The authors thank Dr Jamie Hockin, director, Field Epidemiology Training Program, Health Canada, for his review of this paper.

B. Strauss, RN, BScN, MSc and Mark Bigham, MD, FRCPC

Ms Strauss is a federal field epidemiologist at the Field Epidemiology Training Program with Health Canada in Vancouver. Dr Bigham is a clinical assistant professor in the Department of Health Care and Epidemiology at the University of British Columbia and a physician-epidemiologist at the BC Centre for Disease Control.