Impact of Universal Vaccination of Children on Community Influenza a Mathematical Model

1. Impact of Universal Vaccination of Children on Community Influenza a Mathematical Model Ira M. Longini, Jr. M. Elizabeth Halloran Azhar Nizam Department of Biostatistics Emory University Support: NIAID, NIGMS (MIDAS), CDC

2. This Talk Universal vaccination of children • School children • Preschool children • High risk • Optimal distribution for limited vaccine stocks

3. Vaccine Efficacy Used in the Model: Live and Killed Vaccines • Vaccination before outbreaks Vaccine efficacy • VES = 0.70 – 0.90 (age < 65) • VES = 0.50 (age ≥ 65) • VEI = 0.80

4. Stochastic Simulation Model for Influenza Epidemics

5. Current Model • Basic subpopulations of 2,000 people • All 281 million people in US • Census track level • On average 4,000 people per census track • Influenza natural history

6. Social Structure* • Hierarchy of clusters • Households • Clusters of Households • Preschool playgroups • Schools • Workplaces • Social Setting • Long range travel *Longini, et al. Science, 309, 1083-1087 (2005).

7. Interpandemic Influenza

8. Estimation and Calibration • Illness attack rates estimated using data from published literature • Preschool age: 20 – 30% • Elementary school students: 25 – 35% • Middle school students: 25 – 35% • High school students: 25 – 35% • Young and middle aged adults: 5 – 15% • Older adults: 5 – 10% • Overall: 10 – 20%

9. Current Vaccination Coverage in US • Coverage with mostly killed vaccine • Age 1-18 years: 5.0% • Age 19-64 years: 22.9% • Age ³65 years: 68.1%

10. Average Number of Cases Over Time: Baseline Epidemic in 2000 Person Subpopulation

12. Vaccination of Pre-School and School Children

13. Expected annual influenza illness attack rates in the US, by vaccine coverage levels among young and older children

14. Expected annual number of cases of influenza illness in the US, by vaccine coverage levels among young and older children

15. Expected annual influenza illness attack rates in the US, by vaccine coverage levels among pre-school and school children

16. Simulation Model Comparison to Texas Trial • Simulations run with vaccine coverage similar to past years in Temple • Assume that 41% of MAARI cases actually are infected with influenza* • Indirect effectiveness in adults • VEMAARI • VEadj *Halloran, et al., Am J Epidemiol, 158, 305-311 (2003)

17. Vaccine indirect effectiveness in adultsby vaccine coverage levels among children *1999-2000 and 2000-2001 seasons *20004-2005 season

18. Strategy to reduce influenza-related deaths in the elderly Deaths per 100,000, 65yrs + Current Vaccination Coverage

19. Strategy to reduce influenza-related deaths Deaths per 100,000, 65yrs + Current 20% 5-18yrs Vaccination Coverage

20. Strategy to reduce influenza-related deaths Deaths per 100,000, 65yrs + Current 20% 5-18yrs 50% 5-18yrs Vaccination Coverage

21. Strategy to reduce influenza-related deaths Deaths per 100,000, 65yrs + Current 20% 5-18yrs 50% 5-18yrs 70% 5-18yrs Vaccination Coverage

22. Strategy to reduce influenza-related deaths Deaths per 100,000, 65yrs + Current 20% 5-18yrs 50% 5-18yrs 70% 5-18yrs 90% 65yrs + Vaccination Coverage

23. Pandemic Influenza

24. Optimal Distribution of Vaccine* • Given a limited quantity of vaccine, how should it be distributed among age groups? • Given • The quantity available • The control objective (morbidity, mortality) • Optimization theory • Despite optimal strategy, always give to high risk first *Source: Patel, R., Longini, I.M., Halloran, M.E: Journal of Theoretical Biology (2005).

25. Weights • (presch, sch, ya, ma, oa) • Objective • Morbidity • (1,1,1,1,1) • Mortality per 10,000 illness • (0.263, 0.210, 2.942, 2.942, 199.8)

28. Conclusions and Recommendations • Mass vaccination of school children in concert with vaccination of high risk population • Vaccination of school children is more effective than the vaccination of pre-school children • Pandemic influenza • Determine high transmitting age groups from early epidemiological data • Conduct a nation-wide vaccine trial for mass vaccination of school children: state, regional or city level – Betz Halloran’s idea • Virologic sampling for infection confirmation important • Texas trial is answering this question, but a larger trial would be useful

29. Critical Information Gaps • What is the effect of mass vaccination of school children on a large scale? • Nation wide vaccine trial for mass vaccination of children could augment the Texas trial and help fill in this gap • How does live attenuated vaccine protect compared to killed vaccine? • Non-specific immunity and/or interference • Cross-protective immune responses • How well does influenza vaccine directly reduce transmission to others? • Well designed family and small group trials

30. The End

31. Vaccine Effectiveness

32. Intervention Population* Control Population Overall Vac f Nonvac 1-f Nonvac Direct Indirect Total *Halloran and Struchiner (1991)

33. Multiple Pairs ······· m 1 2

34. Expected annual influenza illness attack rates and overall vaccine effectiveness, by vaccine coverage levels among children VE: 54% (4%, 97%) VEa: 22% (1%, 40%) VE: 63% (15%, 97%) VEa: 26% (6%, 40%) VE: 83% (53%, 98%) VEa: 34% (22%, 40%)

35. Mathematical Formulation • nk – number of people in age group k • ark – illness attack rate in age group k • wk – weight of an illness in age group k • fk – fraction of age group k vaccinated • V – total quantity of vaccine available • (ar1,ar2 , ….. , arK) is a function of (f1,f2 , ….. , fK)