Influenza Vaccine Effectiveness

To the Editor—We read with interest the article by Ray and colleagues and the accompanying editorial by Dr Lipsitch [1, 2]. Ray et al identified a 2-fold increased risk of influenza among individuals vaccinated for >5 months compared to those recently vaccinated. More than a dozen papers have reported findings of declining vaccine effectiveness (VE) with increasing time since vaccination using the test-negative design [3]. We agree with Dr Lipsitch that caution is warranted when inferring that waning immunity is the mechanistic cause of the observed decline in VE. Postvaccination antibody titers decline over time, and although titers usually remain at levels above those associated with protection for the duration of an influenza season [4–6], some degree of decline in VE may result from waning immunity. However, theoretical evidence suggests that an observed decline in VE may be an artifact [7–9]. As discussed by Dr Lipsitch, if influenza vaccine provides protection through a “leaky” mechanism, then vaccination reduces the risk of infection by a certain fraction without rendering complete immunity. Vaccinated individuals may be infected but at lower rates, with the practical implication being that rates of infection become similar between unvaccinated and vaccinated groups with passage of time, manifesting as a decline in observed VE. In short, the decline in VE, or some portion of it, identified by Ray et al and other observational studies could be from artifact and not waning. It is difficult to disentangle waning from artifact related to the leaky vaccine effect. Ray et al used simulation to show that their observed waning effect was unlikely to be artifact, but similar calculations by Dr Lipsitch implied otherwise, suggesting that bias could contribute to the observed decline in VE, although its magnitude would likely be small. Numerous investigators have described the difficulty of elucidating temporal patterns of VE using empirical data due to uncertainty about mechanisms of vaccine protection, potential heterogeneity among vaccinated individuals, choice of VE measure, and other factors [10]. We reiterate the need for development of methodologic approaches in this area and agree with Dr Lipsitch that, in the meantime, multiple sources of information, such as evidence of antibody decline and insights from theoretical expectations and simulation, are needed as corroborating evidence of biologic waning of protection. A vaccine with significant waning of protection would raise important policy questions regarding optimal timing of vaccination campaigns. Delaying vaccination could offer improved protection later in an influenza season but risks lowering overall vaccination coverage and missing opportunities to vaccinate before influenza circulation, leaving individuals unprotected. Given uncertainty about the cause and extent of observed declines in VE and effect of a delay on vaccine coverage, we disagree with premature decisions that may affect vaccine policy, such as the suggestion by Ray et al that vaccination be postponed until November. To best inform policy decisions, a careful vetting of available evidence is required and a clearer understanding of the reasons for observed declines in intraseason VE is necessary—is it waning or is it artifact or some combination of both? Potential conflicts of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.