The influence of incomplete case ascertainment on measures of vaccine efficacy

Background: Motivated by the unexplained variation in the performance of some vaccines across different settings, we extend previous theoretical work to consider the potential impact of incomplete case ascertainment on measures of vaccine efficacy (VE), which is more likely in subclinical or clinically unimportant infections, such as rotavirus gastroenteritis. Methods: By simulating the measurement of VE under outbreak conditions using a discrete time stochastic SIR model, we compare three commonly used measures, VERisk, VERate and VEHazard, calculated respectively based on risk ratio, rate ratio and hazard ratio of disease. We investigate how these measures are influenced by factors such as biological activity, action mechanism of vaccine, proportion of cases ascertained, and underlying force of infection. Results: Under plausibly low levels of ascertainment, the group with the most infections, and therefore the most missed cases, has the most falsely inflated denominator, producing similar rates in the control and intervention groups. As a result, VERare and VEHazard will underestimate the true VE compared to high case ascertainment scenarios. Furthermore, the extent of underestimation is greater for leaky vaccine models with lower biological protective effects and under conditions which are conducive to high transmission. Conclusions: This study demonstrates that a biologically active vaccine may produce a low measured VE under a range of epidemiological, vaccine-related and logistical conditions. Low case ascertainment may partly explain the observed heterogeneity in the performance of rotavirus vaccine across different settings, and should be considered in the design and interpretation of future field trials