The effect of misclassification on evaluating the effectiveness of influenza vaccines

Misclassification is a measurement error and can be considered a type of information bias. Misclassification can occur at both exposure and outcome levels. Nondifferential misclassification causes only a dilution effect leading to underestimation, whereas differential misclassification can have more complicated and serious consequences. To avoid nondifferential diagnosis misclassification, it is necessary to use highly specific diagnostic examinations or criteria such as virus detection to exclude 'false positive' cases, and to limit the observation period to an intensive epidemic period if using less specific diagnostic criteria such as symptoms of influenza-like illness (ILI) or absence from school or workplace. To avoid differential diagnosis misclassification, vaccinated and unvaccinated groups must be equally scrutinized, and such scrutiny is more important than the specificity of diagnosis. So, passive findings from patients with influenza at clinics can cause complicated differential misclassification despite use of highly specific diagnostic procedures because vaccinated and unvaccinated patients may participate differently. Also important is standardization of diagnostic procedure that vaccination anamnesis does not influence diagnosis of influenza, or examination of the influence. Exposure misclassification would mainly underestimate vaccine effectiveness in most situations. Consequently, misclassification of diagnosis, especially differential misclassification, affects evaluation of influenza vaccine effectiveness. (C) 2008 Elsevier Ltd. All rights reserved.