Public Health Impact of Including Two Influenza B Strains in Seasonal Influenza Vaccines

1. Public Health Impact of IncludingTwo Influenza B Strains in SeasonalInfluenza Vaccines Carrie Reed Martin Meltzer Lyn Finelli Anthony Fiore Vaccines and Related Biological Products Committee February 18, 2009 The findings and conclusions in this presentation are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention (CDC)

2. Influenza B viruses • Circulate globally every year • Subdivided into two lineages; currently co-circulate annually • B/Yamagata • B/Victoria • On average, fewer severe complications and deaths than A (H3N2) viruses • Severe complications and deaths caused by influenza B do occur in all age groups • Children appear to have higher rates of infection

3. Influenza B virus lineage and trivalent influenza vaccine • Protection after vaccination with one lineage against other lineage is limited • Co-circulation of both lineages over the past several years means some degree of mismatch between vaccine and circulating strain inevitable

4. Interest in vaccine with strains from both B lineages: Quadrivalent vaccine • Impact due to B strain mismatch: • Reduces overall trivalent vaccine effectiveness • Reduces public confidence in value of influenza vaccine • Increased manufacturing capacity makes quadrivalent vaccine feasible

5. Question • If switch to quadrivalent vaccine (QIV): Compared to trivalent vaccine (TIV), what would be the incremental reduction in cases, hospitalizations, and deaths?

6. Modeling the public health impact • Include last 10 influenza seasons • Characteristic natural variability from season to season • Population average, all ages • May not capture variability by age group • Spreadsheet-based model • User can change inputs • e.g., age-specific data; update to future influenza seasons

7. Step 1: Vaccine Production • For a given production output: Fewer doses of QIV produced than TIV • Question: Over past 10 seasons, how many doses of QIV could have been produced? • Question: How does this relate to the number of doses administered that year?

8. Step 1: What does the model do? • Optimizes the number of doses of a QIV that could be produced with the same production capacity as TIV • Compares the number of QIV doses available to the TIV doses administered

9. Data needed • Annual doses of TIV vaccine produced • Percent reduction in growth of B strain compared to each A strain • Varies by season • Annual doses of TIV administered

10. INPUT: Doses trivalent produced INPUT: Reduced production from A to B viruses Sample page of spreadsheet model with data

11. RESULT: Production and administration of annual influenza vaccine

12. Conclusions, Part 1 • When TIV vaccine supply was similar to demand (e.g., 2002-2005) • Fewer doses of QIV potentially produced than doses of vaccine administered • When TIV supply greatly exceeds demand (e.g., 2005-2008) • Reduced QIV doses available would still exceed the number of vaccines administered

13. Step 2: Public Health Impact • If switch to QIV vaccine: • Over the past 10 influenza seasons; • What would have been the incremental reduction in cases, hospitalizations and deaths?

14. Step 2: What does the model do? • Calculates burden of influenza during each season • Outcomes: rates of illness, hospitalization, death • Compares rates expected with a QIV to rates observed with TIV • Simple model, population average • Can be further refined as data allows

15. Data needed • Rates of influenza-associated health outcomes • Illness, hospitalization, death • By type / subtype / lineage over 10 seasons • Vaccine effectiveness • By type / subtype / lineage over 10 seasons • Virologic surveillance • Annual distribution of type, subtype, and lineage • Vaccine coverage

16. Sample page of spreadsheet model with data INPUT: Rates entered by season

17. RESULTS: Impact of QIV vs. TIV Example 1, 2007-2008 • Total additional outcomes with QIV: • 1,090,514 fewer cases • 7,488 fewer hospitalizations • 321 fewer deaths • TIV supply greatly exceeded demand • No loss in coverage with fewer doses of quadrivalent vaccine • Virologic surveillance: • 29% of virus tested were type B; • 98% were not the lineage in vaccine

18. Impact of QIV varies by season

19. Risk: Fewer doses of QIV? • Potential risk: Decreased coverage • May result in net increase in cases of influenza • From Part 1: In earlier seasons, fewer QIV doses available than administered • Range: 2%-15% fewer vaccinated

20. EXAMPLE 2: 2005-2006 • Virologic surveillance: 19% of circulating strains were influenza B viruses 78% not the lineage in vaccine • Fewer doses of QIV = 8% reduction in persons vaccinated 440,841 fewer cases from improved match with QIV 298,204 increased cases due to decreased coverage • Overall, net decrease of 142,637 cases with QIV

21. EXAMPLE 3: 2004-2005 • Virologic surveillance: 25% of circulating strains were influenza B viruses; 26% not the lineage in vaccine • Problems with production led to decreased supply and administration of vaccine • Loss in coverage if fewer doses of QIV available (15% fewer persons vaccinated) • Net increase of 151,566 cases with QIV

22. Net difference in outcomes with QIV, per US population (Negative numbers indicate net decrease in cases; positive number indicates net increase)

23. Sensitivity analysis:Variability by season and potential reduced coverage

24. Conclusions, Part 2 • When TIV supply was similar to demand (e.g., 2002-2005) • Fewer persons vaccinated with QIV, could have led to modest increases in morbidity or mortality • When TIV supply greatly exceeds demand (e.g., 2005-2008) • Vaccine-induced protection against both B lineages using QIV could have led to modest reduction in morbidity and mortality

25. Limitations • Assumptions from limited data and to simplify model • Data entered as a population average, but may vary by: • Age • Health impact (cases, hospitalizations, deaths) • Strain / lineage • Spreadsheet model can be adapted in future as data is available • Past may not accurately predict the future • Unknown with what frequency different B lineages will circulate • User can change any inputs in the model and recalculate an estimated impact

26. Additional considerations • No economic costs included • No estimate of potential differences in adverse events • No comparison to alternative strategies for reducing influenza impact • e.g., increasing TIV coverage or improving immunogenicity • No consideration of alternating lineage strategy (i.e., alternate B/Yamagata and B/Victoria as vaccine representative)

27. Public Health Impact of Including Two Influenza B Strains in Seasonal Influenza Vaccines Policy Considerations Anthony Fiore Vaccines and Related Biological Products Committee February 18, 2009 The findings and conclusions in this presentation are those of the author and do not necessarily represent the views of the Centers for Disease Control and Prevention (CDC)

28. Production and administration of annual influenza vaccine: era of excess capacity

29. Impact of QIV when capacity greatly exceeds demand

30. Options in an era of excess capacityOption: status quo/no QIV • Focus on improving prevention using TIV • Factors that predict B lineage circulation – improve chances of good match • More immunogenic vaccines • Better understanding of immunologic response to B antigens • Increase vaccine coverage • Pro • Simple • No added expense for immunization program or manufacturer • Con • Infections with influenza B due to lineage not represented in vaccine will continue • Seasons with considerable B activity, mostly due to lineage not represented in vaccine, might occur (e.g., 2007-08) • Excess manufacturing capacity unused

31. Options in an era of excess capacityOption: Move forward on QIV • Pro • Prevention or mitigation of some severe morbidity and mortality associated with influenza B • Public and provider enthusiasm for vaccine that might offer better prevention • Puts excess manufacturing capacity to potential public health benefit • Con • Public health impact of adding 2nd B strain are modest, especially if predominant lineage matches • Increased costs • Immunogenicity data more difficult to interpret for B strains • More data needed

32. Additional data useful for decision-making • Clinical studies • Better understanding of manufacturing issues or constraints • Regulatory path • Economic assessment