Potential effect of virus interference on influenza vaccine effectiveness estimates in test-negative designs.

A hypothetical influenza infection-induced non-specific immunity may reduce the risk of subsequent non-influenza respiratory virus (NIRV) infection and bias the influenza vaccine effectiveness (VE) estimates in test-negative designs (TNDs). We conducted a simulation study using a simple TND model and explored the degree of bias in the VE estimates. The bias was marginal during the usual seasons and most of the time during pandemics; the bias only became large when the influenza infection attack rate increased to pandemic levels (>50%), the true VE was low to moderate, and the non-specific immunity almost completely protected from NIRV infections and lasted at least half the influenza season.