Sir--Rather than clarify the measles, MMR, and autism confusion with your editorial ,¹ you perpetuated the myth that good scientific evidence rejects a link between MMR vaccination and autism.

You quote Taylor and colleagues² as publishing "epidemiological evidence contradicting this alleged association". On March 28, 2000, I presented a talk to the Royal Statistical Society, in which I showed how the currently published data, including that from this study, are consistent with an appreciable number of autism cases being triggered by MMR vaccination. In short, Taylor and colleagues used the wrong study design to detect an association between immunisation and a disease with chronic onset, such as autism. Rather than use a conventional case-control approach, the study used a case-series design. The case-series approach is appropriate for investigating acute adverse events such as febrile convulsion but is not suitable for long-term effects of vaccination.³

Three possible proxy events for onset of autism were chosen: diagnosis, parents' first concern, and regression. Regression was recorded in less than a third of cases and typically occurred 6 months after first concern, diagnosis on the other hand, typically occurred 2 years after first parental concern. The gastrointestinal model for autism described by Wakefield and others,⁴ would suggest a variable delay for perhaps several months between immunisation and first symptoms of autism. It is arguable whether we might expect first parental concern to predate these first symptoms. It is not surprising then that the study found no clustering of diagnosis within 1 year or 2 years of immunisaton, when diagnosis would typically be delayed by more than 2 years. Time of first parental concern and time of regression have highly grouped values at age 12 months, 18 months, and 24 months. This grouping reflects the slow progressive nature of onset of autism and the parent's difficulty in identifying a specific time of onset. That any true association could be spotted when looking at quite short intervals after immunisation, typically 6 months, is unlikely. First parental concern is significantly increased in the 6 months after immunisation (p=0·03). What we can conclude from this study is that either MMR triggering autism is a rare event or, as the model would suggest, it leads to a chronic onset of autistic symptoms.

Like you, I wish to see a full scientific description of O'Leary and colleagues' study as soon as possible. But, we must realise that the current epidemiological evidence does not refute MMR immunisation possibly triggering the onset of autism. In everyone's interest, we must keep an open mind.

J H Roger

Livedata (UK) Ltd, Chinnor Hill Lodge, Chinnor Hill, Chinnor OX9 4BQ, UK (e-mail: james@livedata.co.uk)

1 Editorial. Measles, MMR, and autism: the confusion continues. Lancet 2000; 355: 1379.

2 Taylor B, Miller E, Farrington CP, Cetropoulos MP, Favout-Mayaud JL, Waight P. Autism and measles, mumps, and rubella vaccine: no epidemiological evidence for a causal association. Lancet 1999; 353: 2026-29.

3 Farrington CP, Nash J, Miller E. Case series analysis of adverse reactions to vaccines: a comparative evaluation. Am J Epidemiol 1996; 143: 1165-73 [PubMed].

4 Wakefield AJ, Murch SH, Anthony A, et al. Ileal-lymphoid-nodular hyperplasia, non-specific colitis, and pervasive developmental disorders in children. Lancet 1998; 351: 637-41.