Influenza vaccine for civilian aircrew: is it really effective?

Dear Editor: I feel obliged to comment on the paper “Impact of influenza vaccination on civilian aircrew illness and absenteeism” by Mixeu et al. (5), which does not reflect, in my opinion, your usual high standards. Some of the minor weaknesses include: 1) Some references do not substantiate or do not relate to the argument under discussion: a study by Kumpulainen et al. (4) is cited as supporting the cost-effectiveness of influenza vaccination, whereas its authors clearly suggested the opposite, and Nicholson et al. (7) appears to be cited erroneously; 2) Several recent trials pertaining to the same subject have been completely ignored in discussing the results (1,3,6); 3) The authors did not describe any of the procedures used to randomize or to ensure double-blinding and they did not provide any information on the reasons for withdrawals, although the proportion was significant (27.1%). Most importantly, I am particularly concerned with the sentence in the conclusion (p. 879): “the data indicates that the influenza vaccine is safe and effective in reducing the number, severity, and absenteeism related to influenza-like illness among the aircrew of an international airline”. An examination of Table II of the paper yields the following data for “subjects with at least one episode of influenza-like illness”: Placebo group 98 (out of 299 subjects); Vaccine group 86 (out of 294 subjects); p-value 0.37. Clearly, the influenza vaccine does not show any significant effect when evaluating this outcome. To support their conclusion, the authors have considered data on another outcome: episodes of influenzalike illness, whose reliability in measuring influenza vaccine effectiveness is very poor. In fact, apart from issues concerning clinically confirmed cases as a whole, which are far less accurate than laboratory confirmed cases in assessing influenza vaccine effectiveness (due to the similarities between influenza and many other syndrome symptoms), “cases” instead of “episodes” of influenza are universally preferred as an outcome. Unless two different viruses were circulating simultaneously (which was not stated in the text), it is very unlikely for the same individual to be affected twice during the same epidemic (2). Therefore, the occurrence of two or even three influenzalike illnesses in the same subject during an epidemic is at high risk of being due to an erroneous attribution to influenza of other similar syndromes. Thus, the decision to give more importance to episodes in contrast to cases of influenza may further enhance the likelihood of a severe bias in the diagnostic procedure. Consequently, having based conclusions upon an outcome which is clearly less precise than another one already analyzed is a matter of serious concern, severely weakening the credibility of the overall result. We are currently updating the meta-analysis by Demicheli et al. (2) on the effectiveness of the influenza vaccine for healthy adults. Consistent with Cochrane methodology, we are obliged to include data on the number of subjects with at least one episode as the only outcome in the pooled analysis, obtaining for this study a relative risk of 0.89 (95% CI: 0.70-1.14). This result, clearly, could not support any assumption on vaccine effectiveness in reducing influenza-like illness