In epidemiology, a typical measure of interest is the risk of disease conditional upon exposure. A common source of bias in the estimation of risks and risk ratios is misclassification. Exposure misclassification affects the measurement of exposure, i.e. the variable one conditions on. This article explains how to assess biases under non-differential exposure misclassification when estimating vaccine effectiveness, i.e. the vaccine-induced relative reduction in the risk of disease. The problem can be described in terms of three binary variables: the unobserved true exposure status, the observed but potentially misclassified exposure status, and the observed true disease status. The bias due to exposure misclassification is quantified by the difference between the naïve estimand defined as one minus the risk ratio comparing individuals observed as vaccinated with individuals observed as unvaccinated, and the vaccine effectiveness defined as one minus the risk ratio comparing truly vaccinated with truly unvaccinated. The magnitude of the bias depends on five factors: the risks of disease in the truly vaccinated and the truly unvaccinated, the sensitivity and specificity of exposure measurement, and vaccination coverage. Non-differential exposure misclassification bias is always negative. In practice, if the sensitivity and specificity are known or estimable from external sources, the true risks and the vaccination coverage can be estimated from the observed data and, thus, the estimation of vaccine effectiveness based on the observed risks can be corrected for exposure misclassification. When analysing risks under misclassification, careful consideration of conditional probabilities is crucial.
Bias due to differential and non-differential disease- and exposure misclassification in studies of vaccine effectiveness