The study team reported an extension of an existing static compartmental model to better quantify direct effects of influenza vaccination across multiple seasons. They adjusted the model to permit waning of both natural and vaccine-induced immunity over time and to allow vaccine effectiveness (VE) to differ for hospitalized versus non-hospitalized influenza-associated illness. The authors wrote that these changes enable calculation of deaths averted in addition to cases and hospitalizations prevented. The abstract did not provide detailed numeric values for the waning parameters or the precise compartmental structure beyond this summary. The reader should therefore understand that specific parameter choices and structural equations were not described in the abstract and must be checked in the full manuscript or the repository for reproducibility.
To demonstrate the model, investigators created synthetic datasets representing two contrasting patterns of influenza activity spanning four seasons among older adults (>65 years): a Northern Hemisphere–like seasonal pattern and a hypothetical year-round activity pattern. They applied the same multi-season framework to both datasets. The authors reported outcomes aggregated across the four simulated seasons. The abstract did not specify how seasonality was parameterized or whether stochastic variation between seasons was included; those modeling details appear in supplementary materials or the public code, rather than in the abstract itself.
For the model demonstration the authors assumed VE against symptomatic illness of 37% (17%–53%) and VE against hospitalization of 45% (28%–59%). They also specified two programmatic coverage assumptions. For the Northern Hemisphere–like demonstration they used an average annual vaccine coverage of 33%. For the hypothetical year-round program they used an annual coverage of 20%. The abstract reported these assumed values explicitly. The authors did not, within the abstract, report additional age-stratified coverages, uptake timing windows, or sensitivity analyses beyond the two scenarios summarized.
Data indicated that under the Northern Hemisphere–like activity with 33% average annual coverage, vaccination averted 15.1% (95% uncertainty interval [UI]: 12.2%–16.3%) of influenza-associated illnesses across the four simulated seasons. The annual prevented fraction for hospitalized deaths in that scenario varied by season; the reported range was 14.4% (95% UI: 11.7%–16.8%) to 18.5% (95% UI: 14.9%–21.5%).
Under the hypothetical year-round vaccination program with 20% annual coverage, vaccination averted 4.9% (95% UI: 3.8%–5.4%) of influenza-associated illnesses across the four seasons. The abstract reported annual prevented fractions for hospitalized deaths under that scenario in the 5.5% (95% UI: 4.3%–6.5%) to 5.8% (95% UI: 4.8%–6.6%) range. These outputs illustrate how seasonality, coverage, and differential VE assumptions change the vaccine-averted burden estimated by the model.
The authors noted that the model is implemented in a publicly available tool hosted on GitHub (a link is provided in the source). That repository, according to the abstract, contains the implementation of the described model. The abstract did not enumerate which datasets, calibration scripts, or example scenario files are included; users interested in reproducing or adapting the model should consult the GitHub repository for full code, input files, and any supplementary documentation.
This manuscript is posted as a medRxiv preprint and has not been peer reviewed. The authors declared no competing interests in the source. The abstract summarizes model structure and scenario outputs but omits several methodological details: the abstract did not report specific waning rates, model calibration metrics, sample sizes used for uncertainty quantification, or full sensitivity analyses. Those items were not available in the abstract text and must be consulted in the full paper or repository. Open questions that remain—based on the information presented in the abstract—include the model’s external validation against observed surveillance data, the influence of alternative VE estimates or coverage patterns, and how the model behaves when applied to different age groups or geographic settings. The source did not provide regulatory milestones or planned policy steps; it focused on methodological extensions and demonstration outputs. The final peer-reviewed manuscript, when available, should clarify these unanswered methodological and validation questions.