[Issues associated with evaluating the effectiveness of influenza vaccination by observational studies].

Vaccination against influenza is recommended for subjects at high risk of severe complications if infected with the influenza virus. The effectiveness of influenza vaccination among the elderly has been evaluated in many observational studies but only in a few randomized control trials. However, observational studies have various issues. First, the outcomes should be of high specificity with low nondifferential misclassification, such as laboratory-confirmed influenza or high fever. Differential misclassification between vaccinated and unvaccinated participants may also be a complicating factor due to behavioral differences in the habits of participants with regard to visiting physicians or in the diagnostic processes of physicians. Therefore, preventing differential misclassification by equally precise observation in compared groups seems to be more important than the specificity of outcomes, even if this can be achieved only by techniques with low-specificity outcomes. In observational studies of the elderly, vaccinated elderly individuals may be at low risk of influenza infection or death because they are health conscious. Such confounding may lead to an overestimation of the effectiveness of influenza vaccination. To minimize bias and to adjust for confounders, the effectiveness of influenza vaccination should be compared taking into account the following groups/indicators: influenza season and pre-season, well-matched season and mismatched season between vaccine and circulating viruses, severe season and mild season, high-specificity outcomes and low-specificity outcomes. Low effectiveness is to be expected with the latter of each pair. If no difference in effectiveness is observed between the two conditions in each pair, residual confounding may be present and the true underlying risk factors should be sought. For example, elderly individuals with impaired functional status in addition to chronic diseases may be less able to access vaccination. Hence it is necessary to evaluate the effectiveness of vaccination among such elderly individuals and promote vaccination in those people.