Real-World Effectiveness of Influenza Vaccination Against Medically Attended Influenza and Severe Outcomes Among Older Adults in Yichang, China: A Multi-Season Population-Based Cohort Study

Background/Objectives: Real-world evidence on influenza vaccine effectiveness (VE) against severe outcomes among older adults in China remains limited. We evaluated influenza vaccination against medically attended influenza, influenza hospitalization, and influenza-related intensive care unit (ICU) admission among adults aged ≥60 years in Yichang, China. Methods: We conducted a multi-season population-based retrospective cohort study using linked immunization and healthcare records. Season-specific propensity scores and overlap weighting were used to improve comparability between vaccinated and unvaccinated participants. Weighted Cox regression with individual-level clustered robust variance was used to estimate VE; proportional-hazards assumptions were formally assessed, with extended Cox models fitted when treatment-specific non-proportionality was detected. Results: The cohort comprised 1,106,886 person-seasons, including 31,540 vaccinated and 1,075,346 unvaccinated person-seasons. Vaccination was associated with lower hazards of medically attended influenza (VE, 55.7%; 95% CI, 43.4–65.3%), influenza hospitalization (54.4%; 39.4–65.7%), and influenza-related ICU admission (52.7%; 14.2–73.9%). Protective associations were observed in both winter seasons and persisted when the vaccination window was restricted to 14–180 days before the index date, with VE estimates of 78.4% for medically attended influenza and 83.1% for influenza hospitalization; findings were also robust across other sensitivity analyses. VE point estimates were numerically higher during higher antigenic-concordance periods, although the interaction was not statistically significant. Conclusions: Influenza vaccination within the prespecified pre-index exposure window was associated with lower risks of medically attended influenza and severe influenza-related outcomes among older adults in Yichang. Residual confounding and outcome misclassification cannot be excluded, and the estimates primarily apply to the overlap population rather than all older adults in Yichang.