Skip navigation

MMR and autism: UK database study


Clinical bottom line

A large UK case-control study showed no association between MMR and subsequent development of autism. It also performed a meta-analysis of studies, which overall showed a non-significant reduction of autism rates in children receiving MMR vaccination.


L Smeeth et al. MMR vaccination and pervasive developmental disorders: a case-control study. Lancet 2004 364: 963-969.


Despite considerable research, MMR vaccination of children is still linked to the subsequent development of autism in the some parts of society. This has led to fewer children being protected against measles, and opens the possibility of recurrence of measles epidemics, and the subsequent harm to unvaccinated (and perhaps some vaccinated) children and adults.


The study used the UK General Practice Database. The database is broadly representative of all practices in England and Wales, has been extensively validated, and includes records of immunisation and diagnoses of autism and other pervasive developmental disorders (PDDs). The study included cases of children with a first recorded PDD in the study period (1987 to 2001), and for each case planned five controls. Controls were individually matched by year of birth, sex, and general practice.

Details of vaccination were extracted, together with diagnoses of autism, Aspergenr's syndrome, or other PDD. The date of first diagnosis of PDD was taken as the date of the diagnosis (because final diagnosis could take several months). Duplicate records were excluded, and practices provided anonymised data to help with characterisation of diagnosis in some cases.


There were 1,294 cases and 4,469 controls. In each case 83% were male. For cases, 78% had MMR vaccination at any age, similar to the 82% for controls. For cases 70% had MMR before their third birthday and 62% before 18 months. The median age at first MMR vaccination was 1.2 years for cases and controls.

The mean age at diagnosis was 5.4 years, with 77% of diagnoses being for autism and 23% for other PDDs. The age at which a PDD was diagnosed was the same in both groups (Figure 1). Most children were diagnosed after their fifth year.

Figure 1: Age of first diagnosis of PPD for vaccinated and unvaccinated children

The odds ratio for the association between MMR vaccination before the index date and a diagnosis of PDD was 0.73 (95% CI 0.59 to 0.91). After adjustment for age at which children joined the database it was 0.86 (0.68 to 1.09). Analyses based on age of children when first vaccinated (cut offs of three years and 18 months) produced similar results. Results were also similar for autism and other PDDs analyses separately.

Systematic review and meta-analysis

The paper also sought other studies that assessed the risk of PDD in those who had MMR vaccine and those who did not. Eligible studies were those with an overall effect measure could be obtained. Overall there were three such studies, including the one reported above. One small study using the UK GP database had to be excluded because it would have used the same sample.

The results (Figure 2) showed that risk was not significantly different, with an overall relative risk of 0.87 (0.76 to 1.001).

Figure 2: Meta analysis of risk of developing PDD with MMR vaccination


The UK GP database study was impeccable. It showed no increased risk of autism or PDD with MMR, and looked at a whole range of different possible analyses to investigate whether there could be associations with age at vaccination, or type of condition. There were none.

Moreover, it also showed that while age of vaccination was 1.2 years, the age of first diagnosis was 5.4 years, a four-year gap. This is quite different from the media presentation in which a perfectly normal child begins to have behavioural problems within weeks of MMR vaccination.

And the systematic review and meta-analysis was within a whisker of having a statistically significant result that went the other way, that there was a reduced risk of autism in children who had been vaccinated with MMR. Perhaps it is too early to say, but another study like the ones we have so far would nail it down. In that case parents who did not have their children vaccinated with MMR would put them at risk of both harm from measles, and harm from being at increased risk of developing autism.