Toward a better understanding of the relationship between influenza vaccine efficacy against specific and non-specific endpoints and vaccine efficacy against influenza infection

Abstract

In influenza vaccination studies assessing vaccine efficacy (VE), both specific and non-specific endpoints (outcomes) are used. We present a formula for the relationship between VE against influenza-related outcomes (VE_O), specific and non-specific, and that against influenza infection (VE_I). In its simplest form, the formula comprises two additional parameters: the influenza attack rate among unvaccinated subjects, and the relative risk of the outcome for influenza infected subjects versus non-infected subjects. Both parameters may show large between-seasonal variation, which translates to a large between-seasonal variation of  VE_O estimates. With the full form of the formula it can be shown that, contrary to popular believe, VE_O may be greater than VE_I. We argue that interpreting VE_O estimates in terms of “low” or “high” is not possible without taking the costs of an outcome case into account. We conclude that the decision to use a non-specific endpoint as surrogate for influenza infection should be taken in the awareness of these limitations.