Modeling potential herd immunity from influenza vaccination

1
Modeling potential herd immunity from influenza
vaccination

• Martin Meltzer, Ph.D.
• Office of Surveillance, NCID

2
Why how model?

• Lack of data
• No statistical analyses all questions
• Want extrapolations, interpolations
• Answer what ifs
• Many methods available
• Deterministic fixed relationships
• Stochastic probabilities define relationships
• Different levels possible

3
Papers considered One Model

• Elveback et al Am J Epidemiol. 1976 103153
• Longini et al Vaccine, 2000181902-1909
• Halloran et al. Vaccine 2002 20 3254-3262
• Weycker et al Vaccine 2005231284-1293
• Background
• Longini et al. Internat J Epi 198413496-501
• Longini et al Am J Epi 1982115736-751.

4
The model(Halloran et al Vaccine 2002 20
3254-3262)

• 2000 people age/ gender as per census
• Households in 4 neighborhoods
• 1 HS, 1 MS, 2 ES, small or large playgroups
• No risk groups by pre-existing conditions
• No nursing homes
• Weyker et al. bigger 10,000 in 5 neighborhoods
• Individual agent level model
• Start Randomly allocate 12 infectives
• Further introductions not addressed

5
Transmission data Sources

• Household transmission Halloran et al
• Tecumseh 1977/78 and 1980/81
• (Addy et al., Biometrics 199147961-974)
• Other rates . . chosen to calibrate . . . to
  certain illness attack rates in each age group.
• Young children 27 (95 CI 2, 42)
• Older children 29 (95 CI 3, 40)
• Adults (all) 10 (95 CI 2, 14)
• OVERALL 15 (95 CI 2, 22)
• Wyecker et al average 11 - sensitivity
  analysis
• Elveback et al 1957 and 1968 pandemics

6
Who infects whom Matrix of transmission
probabilities
Table 7 Transmission probabilities in
simulated populations Contact group Children
Adults Pre-school School Younger
Older Small play group Large daycare
Elementary Middle High Small play groups
0.0680 Large daycare centers
0.0275 Elementary school
0.00577 Middle school
0.00477 High school
0.00177 Family Child 0.08
0.08 0.08 0.08 0.08 0.03 0.03 Adult
0.03 0.03 0.03 0.03 0.03 0.04 0.04
Neighborhood 0.00001 0.00001 0.000027
0.000027 0.000027 0.00012 0.00012 Community
0.000005 0.000005 0.000015 0.000015
0.000015 0.000115 0.000115
Source Halloran et al. Vaccine, 2002203254-3262
7
Who infects whom Close up of some daily
transmission probabilities
Source Halloran et al. Vaccine, 2002203254-3262
8
Results Cases averted
Base vaccinations 5 children 22.9 of
19-64yrs 68.1 of 65yrs
Source Weycker et al., Vaccine 2005 231284-1293
9
Results EffectivenessMean (S.D.)
Note Adults vaccinated 22.9 of 19-64yrs 68.1
of 65yrs
Source Halloran et al., Vaccine 2002
203254-3262
10
Critical fraction (f) vs. Ro Efficacy
(susceptible) 92
0.8
0
Efficacy infectiousness
0.7
80
0.6
Critical fraction f
children vaccinated to stop epidemic
0.5
Source Longini et al Vaccine, 2000 18 1902-1909
0.4
0.3
1.50
2.00
3.00
2.50
Ro
11
Some evidence of vaccine-generated herd immunity
Start Japanese vaccination of children
Source Reichert et al. N Eng J Med
2001344889-896
12
Seen pattern before? Infectious Disease Mortality
in the United States, 1900 to 1996
Source Armstrong et al, JAMA, 1999 28161-66
13
Background to Japanese experience Changes in
Population Pyramid, Japan, 1950 - 2000
Male
Female
Male
Female
Age
Population (thousands)
Population (thousands)
Did such changes alter the transmission matrix in
Japan? Account for such changes in U.S. (when
data sources are old)?
14
Logistics, logistics, logistics
Assume 80 targeted prophylaxis with anti-viral
drugs of contacts of identified cases, for up to
8 weeks
Source Longini et al., Am J Epidemiol,
2004159623-633
15
Pandemic vaccination progress toward meeting 80
goal for target groups 5 million doses
available per week
1 dose/ person
2 dose/ person
High Risk Pop. Medical Care 92 million
Vaccinated Population (cumulative millions)
High Risk Pop. Only 85 million
Medical Care Workers Only 13.5 million
Also, roughly equal to immunizing 80 of all
children under the age of 18 Assume no waste
and all vaccine goes to 80 of target group
16
3 major concerns

• Confidence intervals
• How wide? How big a gamble?
• Available data indicative of pandemic?
• Transmission matrix realistic?
• Logistics a BIG problem
• Delay of 1 -3 months in vaccinating target group?
• Could it actually be done production delays?

17
Appendix
18
The model Part II

• Pre-intervention vaccination rates
• 5 child 22.9 in 19-64yrs 68.1 in 65yrs
• Represents 64.5 million doses adults gt 18 yrs.
• Vaccine efficacy
• Protective (susceptible) efficacy 70
• Infectiousness efficacy 80
• Intervention Vaccinate children 1-18yrs
• 30 or 50 or 70

19
The model Part III

• Latent period 1-3 days, mean 1.9
• Infectious period 3-6 days, mean 4.1
• Prob. infective is symptomatic 67
• Asymptomatic 50 are infectious
• Prob. of withdraw, given symptoms
• Pre-school 80 school kid 75 adults 50
• Withdraw on days 0 through 2

20
More ResultsWeycker et al Vaccine 2005 23
1284-1293

• Attack rates
• 0-4 yrs 17
• 5-18 yrs 24
• 19 yrs 7
• Overall average 11
• Vaccine efficacy
• 70 susceptibility children younger adults
• 50 susceptibility older adults
• 80 infectiousness all ages
• Sensitivity analyses
• Reduced within house transmission child-to-adult
  by 50 increased adult-to-adult

21
Summary and Conclusions

• One model/ community all U.S.?
• Transmission matrix
• Hold for all years and all communities?
• Some other inputs also kept fixed
• Outside introductions not explored
• Only sub-divide by age (no risk)
• Vaccinate all children 6 months 18 yrs
• Vaccinate other age groups 64.5 mil doses
• Cost of vaccinating children not explored
• In U.S., approx. 77 million children aged lt1 18
  yrs
• Models demonstrations, not absolute proof

22
The model Transmission

• Each person, each day, prob. of contact
• Pre-school (2 types) School ( 3 types)
• Family (child adult)
• Neighborhood Community
• Calculate probability of transmission