Online Program

Return to main conference page
Wednesday, September 25
Wed, Sep 25, 9:45 AM - 10:30 AM
Marriott Foyer
Poster Session

Applying Statistical Methods for Improved Influenza Vaccine Formulation (300832)

*Caroline Kaminski Cunningham, Washington-Lee High School 

Keywords: influenza vaccine, vaccine effectiveness, multivariate regression analysis

Influenza vaccine effectiveness depends on the accuracy of the annual vaccine formulation. The methodology of influenza vaccine formulation has not yet been perfected; in recent years, the vaccine effectiveness has been as low as 36%.

This project applied statistical analyses to determine the relationship between factors in the influenza vaccine and vaccine effectiveness with the ultimate goal being to improve influenza vaccine formulation and development. First, the impact of the accuracy of each section of strains in the influenza vaccine (A H1N1, A H3N2, and B) on influenza vaccine effectiveness was calculated. It was determined that the accuracy of the influenza B strains have the most substantial impact on the overall vaccine effectiveness (p = 0.032). This result suggests that the primary focus in vaccine formulation should be placed on choosing the influenza B strain(s).

Second, the influence of the glycoprotein ratio on the influenza vaccine effectiveness was studied. The glycoprotein ratios studied were: hemagglutinin accuracy greater than neuraminidase accuracy, hemagglutinin accuracy equal to neuraminidase accuracy, and hemagglutinin accuracy lower than neuraminidase accuracy. Having an equal hemagglutinin and neuraminidase accuracy ratio was most advantageous, with the highest mean vaccine effectiveness and lowest standard deviation (p = 0.026). Additionally, the percentage of A H3N2 influenza strains subtyped, which the vaccine offers the least protection against, decreases with an equal glycoprotein ratio (p < 1/1000). These findings, therefore, support the standardization of the neuraminidase component of the influenza vaccine.

Finally, using multi-regression analysis, an equation was developed to predict vaccine effectiveness for both hemispheres (R-squared = 0.66). This equation has applications both for countries without quality influenza surveillance and for estimating vaccine effectiveness before influenza vaccines are administered.