Farrington, C. Paddy and Miller, Elizabeth
Informed choice, balance and the MMR saga.
The Lancet Infectious Diseases, 5 pp. 2–3.
Tom Jefferson states that the three studies he and co-workers included in their review of the evidence on measles, mumps, and rubella (MMR) vaccination and autism contained “fundamental methodological weaknesses”, primarily because the studies did not include an unvaccinated control group. He describes such studies as “uncontrolled” and “always prone to the interference of known and unknown biases and confounders”. He goes on to question the epidemiological and statistical competence of those who published these studies without warning of their “methodological limitations”. We were curious to learn on what evidence he bases these damning statements. On further inspection, one of the three studies he quotes turns out to be a case-control study on the association between MMR and inflammatory bowel disease, not autism, and is therefore not relevant. Conversely, his review ignored several other studies relevant to the association between MMR vaccination and autism. We refute his claims that the two studies he did consider on MMR and autism are uncontrolled and biased. Both studies are controlled, with exposure defined in terms of time since vaccination, and both studies also included unvaccinated children. Studies of this type, incorrectly described by Jefferson as “time series”, or “before and after” studies, have been used for many years in vaccine safety assessment. There is an ample literature on them, including several large cohort and self-controlled case series studies on MMR, which were unaccountably excluded from Jefferson's review (for example, Griffen et al and Farrington et al). The self-controlled case series method is self-matched, and hence less prone to confounding bias than either case-control or cohort studies, for which adjustment is often either not possible, or achieved only imperfectly through proxy variables. Jefferson argues that “every scrap of knowledge” must be put to good use in assessing vaccine safety. We agree. However, his assessment of the epidemiological evidence of MMR involvement in autism, and its safety more widely, was based on studies selected through inappropriately restrictive inclusion criteria, incoherently applied, and subjectively assessed. Such a distorted view of the evidence cannot constitute an appropriate basis on which to draw wider lessons about vaccine safety. In 2002–03 CPF was an expert witness in litigation on MMR and autism, for which he was instructed by the defendants. Since 2003 his department is in receipt of an EPSRC CASE studentship partly funded by GlaxoSmithKline Biologicals. EM's department has received funding from vaccine manufacturers for carrying out clinical trials and for provision of surveillance reports.
Actions (login may be required)