1. Cost-effectiveness of paediatric seasonal influenza vaccination in England and Wales
RJ Pitman ICON Health Economics, Oxford, UK Canadian Public Health Association Conference Toronto, Canada May, 2014

2. Why conduct cost-effectiveness analyses?
Health policy is formulated to maximise the health of the population
Budgets are finite
So how do we most efficiently use these limited resources?

3. Incremental costs and benefits
Outcomes
Policy A
 Cost
 Benefit
Choice : Policy
A or B ?
Costs
Outcomes
Policy B
Does the extra benefit justify the extra cost ?

4. Cost-effectiveness plane
New treatment more costly
New treatment more costly and more effective
Old treatment dominates
New treatment less effective
New treatment more effective
New treatment less costly and less effective
New treatment dominates
New treatment less costly

5. Cost-effectiveness and opportunity cost
QALYs gained
Cost
Cost-effectiveness threshold
£20,000 per QALY
£60,000
Price > P*
£30,000
per QALY
£40,000
Price = P*
£20,000
per QALY 2
£20,000
Price < P*
£10,000
per QALY 1
Net Health Benefit
1 QALY 3
Net Health Benefit
-1 QALY

6. The region of cost-effectiveness
Difference in cost
Cost-effectiveness threshold
Difference in effect
Region of cost-effectiveness

7. Cost-effectiveness plane
l = cost-effectiveness threshold l

8. Cost effectiveness acceptability curve
New Treatment
100%
80%
Probability most cost-effective
50%
20%
Current Treatment
Cost-effectiveness threshold (l)

9. Moving from the ICER to net benefit
Threshold defines value of health outcome
Standard ICER decision rule:
ΔC/ΔQ < λ
Where λ is the threshold
Net monetary benefit:
(ΔQ x λ) – ΔC > 0
Net health benefit:
ΔQ – (ΔC/ λ) > 0

11. Influenza vaccination
In England and Wales vaccination against seasonal influenza was only offered if:
Over six months old and at an increased risk of influenza complications
This included everyone over the age of 65 years and anyone younger who falls into a predefined risk group

12. Influenza vaccination
Base case
Influenza vaccination (status quo)
6mo - <65yrs at risk of influenza related complications
65+yrs vs
Paediatric influenza vaccination
Status quo + 2yrs - <19yrs

13. The components of a vaccine cost-effectiveness model
Pitman et al. Vaccine (2012) 30:

14. Who Acquires Infection From Whom (WAIFW) matrix
Mixing is heterogeneous
Mossong J et al. PLoS Med (2008) 5:

15. The components of a communicable disease model
Vynnycky E et al. Vaccine (2008) 26:

16. Primary care physician visits
Deaths
Hospitalization
Outpatient visits
Primary care physician visits
Symptomatic disease
Influenza infections
AH1N1
AH3N2
B(Yam)
B(Vic)
Pitman et al. J Infect (2007) 54:

17. Incremental cost effectiveness analysis over 15 years, at 50% uptake
Policy option
QALYs lost (millions)
Cost (millions)
Incremental QALYs (millions)
Incremental Cost (millions)
ICER
No Vaccination
10.25
£8,123
Dominated
Current policy
2.43
£6,730
7.82
-£1,393
Cost saving 
Current policy + vaccination in year olds
1.75
£6,759
0.69
£29
£43
Current policy + vaccination
in year olds
0.20
£7,870
1.55
£1,111
£719

18. 15 year cumulative burden of influenza, at 50% uptake
No vaccination
CP yrs
CP yrs
Current practice (CP)
Pitman et al. Vaccine (2013) 31:

19. Cost effectiveness acceptability curves, using 15 year cumulative data, at 50%
No vaccination
Current policy
CP yrs
CP yrs
Pitman et al. Vaccine (2013) 31:

20. Conclusions
Paediatric vaccination is likely to result in substantial health savings both within and beyond the vaccinated cohorts of children
While there is considerable uncertainty in the system, annual paediatric influenza vaccination was consistently estimated to be cost-effective

21. Thank You Contributors: Lisa Nagy – ICON Health Economics
Mark Sculpher – York Centre for Health Economics

22. Reserve slides

23. Decisions when a treatment does not dominate
Assessing the opportunity cost
What existing treatments will have to be displaced?
What health benefits will be forgone?
A rule of thumb
How does the extra cost of a unit of benefit compare with previous decisions?
What is society willing to pay for an extra unit of benefit?
Increase insurance premiums/taxation to provide new intervention

24. Influenza virus Constantly changing virus Comes in two types: A and B
Influenza A
Responsible for pandemics
Has numerous subtypes
A(H1N1) – 1918
A(H2N2) – 1957
A(H3N2) – 1968
A(H1N1/09) – 2009
PHIL ID #11823, Dan Higgins, CDC

25. Burden of influenza in England & Wales
The annual health burden of seasonal influenza is considerable
Very roughly:
1 million General Practice consultations
25,000 Hospitalisations
20,000 Deaths
Pitman R et al, J Infect, 2007, 54,

26. Estimated annual rate of influenza A related primary care consultations
Pitman R et al, J Infect, 2007, 54,

27. Estimated annual rate of influenza A related hospitalisations
Pitman R et al, J Infect, 2007, 54,

28. Estimated annual rate of influenza A related deaths
Pitman R et al, J Infect, 2007, 54,

29. The debate: Who should we be targeting?
Those at greatest risk of complications?
Or a combination of both?
Those most likely to transmit the virus?
Influenza virion TEM: #10073 CDC/ Dr. Erskine. L. Palmer; Dr. M. L. Martin

30. Objective
To assess the cost effectiveness of adopting a policy of routine childhood influenza vaccination in the England and Wales, taking account of the dynamics of transmission and indirect protection (herd protection)

31. Herd Immunity Susceptible Immune following infection Vaccinated
Protected by herd immunity

32. Current policy High risk groups 65+ years old
6 months – 64 years old in the following groups:
Residents of nursing or residential homes for the elderly and other long-stay facilities
or with the following conditions:
Chronic respiratory disease (includes asthma treated with continuous or repeated use of inhaled or systemic corticosteroids or asthma with previous exacerbations requiring hospital admission)
Chronic heart, liver or renal disease
Chronic neurological disease
Diabetes mellitus
Immunosuppression because of disease or treatment
HIV infection (regardless of immune status)
Carers of the above
Health care workers
Pregnant women

33. Methods: Univariate sensitivity analysis
Population mixing
Homogeneous (random)
Basic reproductive rate
1.4
2.2
No Seasonality
Seeding
10 / year / age band; 5 – 50 years of age
Vaccine Coverage
10%
50%
80%

34. % uptake in total population2
Status quo
Age groups in model
% efficacy 1
% uptake in total population2
0 - <1
60%
0.1%
1 - <2
2 - <5
1.4%
5 - <11
11 - <19
19 - <50
75%
5.6%
50 - <65
65+
50%
73.5%
Elderly (65+yrs) and at risk groups vaccinated
Annually from 2000 onwards
Jefferson T. et al Cochrane Database Syst Rev, CD
Jefferson T. et al Cochrane Database Syst Rev, CD
Rivetti, D. et al Cochrane Database Syst Rev, CD
2. Health Protection Agency data

35. Paediatric vaccination
Annually from 2009 onwards
Efficacy 80% 1,2
Coverage 50%
Target age groups
Pre-school: 2 - <5
Primary School age: 5 - <11
Secondary school age: 11-<19
Jefferson T. et al Cochrane Database Syst Rev, CD
Jefferson T. et al Cochrane Database Syst Rev, CD
Rivetti, D. et al Cochrane Database Syst Rev, CD
Belshe, R. B. et al N Engl J Med, 2007, 356,
Rhorer, J. et al Vaccine, 2009, 27,

36. Cost-effectiveness analysis

37. Assumptions Health Service (NHS) perspective
3.5% discount rate applied to both costs and benefits

38. Costs
Vaccination
GP consultation
Vaccine price point
TIV – mean list price
LAIV = TIV
Health outcomes – age stratified mean cost of influenza related
Hospitalisation
Costs inflated to 2009 prices, where appropriate

39. Outcomes Averted Quality adjusted life years
General practice consultations
Hospitalisations
Deaths
Quality adjusted life years
QALY decrements based on estimates from 2003 HTA report1
Adult QALYs used
Turner D et al Health Technol Assess. 2003; 7: 1-170

40. Sensitivity analyses Univariate sensitivity analysis
Probabilistic Sensitivity analysis
Variation in probabilities: beta distribution
Variation in costs and utility decrements: gamma distribution
Probability cost-effective: Cost-effectiveness acceptability curve (CEAC)
Probability of option with highest net benefit being cost-effective: Cost-effectiveness acceptability frontier (CEAF)
Extreme value analysis

41. Extreme Value Analysis
The transmission coefficient (R0 of 1.4, 1.8 and 2.2)
Infectious cases seeded into the population each year (50, 100, 150)
Duration of natural immunity (influenza A: 5 years, 6 years, 7 years; influenza B: 11 years, 12 years, 13 years)
Duration of infectiousness (1 day, 2 days, 5 days)
Percentage of infected individuals that experience symptoms (55%, 64%, 73%)
Latent period (1 day, 2 days, 3 days)
Duration of vaccine induced immunity (Flu A: 3 years, 6 years, 7 years; Flu B: 6 years, 12 years, 13 years)

42. Simulated influenza
Vaccination of elderly
Paediatric vaccination

43. Results: Sensitivity of the 15 year cumulative averted influenza cases per 100,000 population, assuming 80% coverage of 2 to 18 year old children with LAIV in addition to current practice

44. Extreme value analysis
Influenza A
Influenza B

45. The cost of an influenza related GP consultation
Age Group
GP consultations
0-11 mo
£88
12-23 mo
24-59 mo
£65
5-10 years
£54
11-18 years
£66
19-49 years
£85
50-64 years
£101
65+ years
£100
General Practice Research Database
Personal and Social Services Research Group Report 2008

46. The cost of an influenza related hospitalisation
Median
Median Cost
Length
of Stay yr 6
1,606 yr
1,634 yr
1,662 yr 7
1,983
65+ yr 19
5,354
All Ages 8
2,123
Hospital Episode Statistics
NHS National Schedule of Reference Costs

47. Annual primary care and hospitalisation costs of influenza in England & Wales
Total annual cost ~ £175,000,000
Hospital Episode Statistics
NHS National Schedule of Reference Costs
General Practice Research Database
Person and Social Services Research Unit annual report
Pitman R et al, J Infect, 2007, 54,

48. Results: Sensitivity of ICER to time horizon, at 50% uptake

49. 15 year estimated cumulative burden of influenza, at increasing levels of paediatric vaccine uptake

50. Cost effectiveness acceptability frontier, based on 15 year cumulative data, at 50% uptake

51. Limitations
Model incidence calibrated against Tecumseh data (1970s, US data)
It was not possible to benchmark the model simulations against actual European data on influenza virus infection rates. Obtaining such sero-epidemiological data will help reduce uncertainty within the parameter estimates
High level of uncertainty in many of the parameters
Vaccination behaviour
Simple treatment of antigenic drift
Cross-protection

52. Is there evidence for herd immunity?

53. Tecumseh, Michigan, USA
1968
Hong Kong Flu
AH3N2

54. Tecumseh study Tecumseh Adrian Population ~10,000
Respiratory illness surveillance project since 1965
~360 families followed at any one time
60% random selection
40% selected with chronic respiratory disease & matched controls
Followed for one year then replaced with another family
Vaccinated 85.8% school age children (Elementary through to High School)
Monovalent vaccine (A2/Aichi/2/68) vs. AH3N2 virus (Hong Kong Flu)
School absenteeism monitored along with respiratory illness surveillance
Population ~20,000
12 miles south of Tecumseh
Significant population interchange
Weekly family surveillance programme in place
150 households
Each family followed for 3 weeks
Same questions on respiratory illness as in Tecumseh
School absence data also collected
Monto AS et al, Bull World Health Organ, 1969, 41,

55. Tecumseh study
Two weeks after vaccination began the first isolate of AH3N2 was obtained in Tecumseh

56. School absenteeism 1968 – ‘69
Monto AS et al, Bull World Health Organ, 1969, 41,

57. Incidence of respiratory illness
Monto AS et al, Bull World Health Organ, 1969, 41,

58. Age specific weekly mean rate of respiratory illness
85.8% vaccine coverage
Monto AS et al, J Infect Dis, 1970, 122, 16-25

59. Japan – paediatric influenza vaccination
1962 – 1987 mandatory influenza vaccination of all schoolchildren
Law relaxed in 1987
Repealed in 1994
Thomas Reichert analysed excess mortality
Reichert et al. N Eng J Med 2001;344:

60. Excess deaths attributed to pneumonia and influenza – 5 year moving average
Reichert et al. N Eng J Med 2001;344:

61. Population pyramid, Japan, 1950 - 2000
Age
Male
Female
1950
2000
Population (thousands)

62. Herd immunity demonstrated in randomised control trial in 2010
Loeb et al. JAMA 2010; 303:

64. A simple epidemic schematic with replenishment of susceptible individuals
Effective reproductive rate 2 1
Effective reproductive rate
= R0
Infections
Time
Epidemic curve

65. Communicable disease epidemiology
Effective reproductive rate 2 1
Immune
Infections
Vaccination
Time
Epidemic curve
Effective reproductive rate

66. Communicable disease epidemiology
Effective reproductive rate 2 1 V
Immune
Infections
Time
Epidemic curve
Effective reproductive rate
Vaccination

67. Communicable disease epidemiology
Effective reproductive rate 2 1 V
Immune
Infections
Time
Epidemic curve
Effective reproductive rate
Vaccination

68. Communicable disease epidemiology
Effective reproductive rate 2 1 V
Immune
Infections
Time
Epidemic curve
Effective reproductive rate
Vaccination

69. Communicable disease epidemiology
Effective reproductive rate 2 1 V
Immune
Infections
Time
Epidemic curve
Effective reproductive rate
Vaccination

70. Communicable disease epidemiology
Effective reproductive rate 2 1 V
Immune
Infections
Time
Epidemic curve
Effective reproductive rate
Vaccination

71. Replenishment of susceptibles
Generation of immunity
Rate of viral spread R0
Viral generation time
Vaccination
Coverage
Frequency
Behaviour
Loss of effective immunity
Waning immunity
Antigenic drift / shift
Births

72. QALYs “Generic” measure of health Combines length and quality of life
Applicable to a wide range of clinical areas
Combines length and quality of life
Utilities used as quality weights:
Survival = 10 years
Utility score = 0.5
QALYs = 10 x 0.5 = 5 QALYs

73. Calculation Of QALYs UTILITY 0.9 3.6 QALYs (With treatment)
2.6 QALYs gained
0.5
1 QALY
(No treatment) 1 2 3 4
YEARS

74. Calculation Of QALYs UTILITY 0.9 With Treatment QALYs Gained 0.5
No Treatment 5 10 15
YEARS

75. A QALY Is A QALY
UTILITY 1
0.5
0.3 1 2 3
YEARS