Author's response to influenza vaccination: policy v evidence

BMJ 2006; 333 doi: http://dx.doi.org/10.1136/bmj.39045.403808.1F (Published 30 November 2006) Cite this as: BMJ 2006;333:1172
  1. Tom Jefferson, coordinator
  1. 1Cochrane Vaccines Field, Anguillara Sabazia, Roma 00061, Italy [email protected]

    My analysis was based on 206 studies (several million observations' worth of data) included in systematic reviews spanning some 40 years. The hypotheses by Mandl do not fit some of the evidence in the elderly population.1 He cannot explain how in years of good matching between vaccine antigenic content and circulating viruses the vaccines fail to prevent deaths from all respiratory diseases in elderly community dwellers (1.32, 95% confidence interval 1.25 to 1.39, 426 668 observations) while at the same time preventing 42% (25% to 55%, 404 759 observations) of deaths from all causes,2 presumably including deaths from falls, accidental poisoning, accidents, hypothermia, and so on.

    Fedson and Nichol deride my choice of example of poor methodological quality of a large number of available cohort studies.1 The authors of the studies either did not know such details or like Fedson and Nichol thought them irrelevant and would leave a reader to work them out from “official records.” Vaccine matching and level of circulating influenza viruses are the most important predictors of vaccine efficacy and effectiveness. The closer the match and the higher the viral circulation, the better the performance of the vaccine.3 Without such knowledge it would be very difficult to give an honest and reliable assessment of the effects of the vaccine. That is one of the reasons why these studies are of poor quality.

    I note with worry their statement that decisions should be made on three of the most notoriously biased sources of information: non-randomised studies, expert opinion, and economic evaluations.4 5 It is precisely because most comparative evidence on elderly people comes from non-randomised studies that we are left with the question: are the effects we witness due to the vaccines or are they due to confounding? The tone of the response by Fedon and Nichol (lack of vaccines' effect in small children is undoubtedly due to small numbers and my concern over lack of vaccine safety data—a statement from which they omitted the key word “comparative”) implies that my review seemed to be questioning a dogma. Heretics like me get short shrift.

    I repeat my statement that especially in elderly people, an insufficient number of field trials has been conducted (five, of which only one has been carried out in the past decade) to allow reasonable certainty of the effects of inactivated vaccines. The nature of the evidence from non-randomised designs when analysed critically and exhaustively is weak and contradictory. I repeat my observation that the totality of safety evidence from comparative sources (studies in which a proportion of participants were contemporaneously exposed or not to the vaccines) is tiny in small children (35 observations) and small in the elderly (2963 observations).


    • Competing interests: TJ owned shares in Glaxo SmithKline and received consultancy fees from Sanofi-Synthelabo (2002) and Roche (1997-1999).


    View Abstract