Predicting influenza vaccine-elicited antibody responses with practical point systems

Influenza vaccination plays a crucial role in reducing morbidity and mortality from influenza. However, its effectiveness varies due to multiple factors. Reliable point systems combining age, sex, BMI, vaccination history, and other baseline characteristics could aid in making evidence-based decisions regarding influenza vaccination.

Using human vaccination cohort data from the University of Georgia (UGA) over multiple influenza seasons, we developed two point systems: the Simple-Test score (STS) and the No-Test score (NTS). These scores predict vaccine-elicited antibody responses measured by hemagglutination inhibition (HAI) titers. Data from four influenza seasons (2016-2017 to 2019-2020) were used for model development and validation.

The STS and NTS demonstrated good performance in discriminating between predicted lower-, moderate-, and higher-response groups. The AUC values for the STS were 0.943 for derivation and 0.841, 0.936, and 0.796 from the validation cohorts for 2016-2017, 2018-2019, and 2019-2020, respectively. Age, race, BMI, baseline HAI titers, and vaccination history significantly influenced the point system's performance. The point system showed robustness across age groups (teenagers, adults, and elderly). The AUC values for the NTS were 0.913 for derivation and 0.658 to 0.875 for validation datasets.

We successfully developed and validated two practical point systems to predict individual-level influenza vaccine-elicited antibody responses. These systems could facilitate personalized vaccination recommendations, policymaking, and resource allocation in influenza vaccination programs. The proposed point system is also a valuable tool for targeting populations that are likely to benefit most from influenza vaccination.