Health economic evaluation of vaccine: the example of varicella-zoster virus

Transcription

1 Health economic evaluation of vaccine: the example of varicella-zoster virus Benoît DERVAUX, CNRS, Catholic University of Lille Advances in infectious diseases modelling Annecy, Les Pensières, December 10-12, 2007

2 Resource allocation in health care The Fourth hurdle to market: Safety Efficacy Quality Efficiency Economic evaluation = the «value for money» of new interventions

3 Types of economic evaluation How health consequences are valued? Cost-effectiveness analysis: natural units Cost-utility analysis: QALYs Cost-benefit analysis: in monetary terms Ordered evaluations: CBA > CUA > CEA

4 Are vaccines different? (1) Vaccination has direct and indirect effects. Such externalities may be positive or negative (herd immunity) Epidemiological impacts of vaccination usually occur in the long term (sensitivity of the results to discounting rules) IMPORTANCE OF MODELING Many vaccines prevent short-lived illness in very young children (QoL measurement, valuation of indirect costs) Some infections are eradicable

5 Are vaccines different? (2) Key elements: Herd immunity, externalities Discounting, choice of end-points and time windows: prevention vs cure Risk of underestimation of the value of vaccines

6 Discounting Should the health consequences be discounted? Do time preferences depend on decision time horizon? Should we use exponential or (quasi)- hyperbolic discounting models? What is the real objective of preventive programs? Should we consider risk reduction or long-term consequences on morbi/mortality as endpoints of vaccination programs? Are time preferences identical for all commodities? Should costs and health consequences be discounted at the same rate?

7 About varicella 90% of varicella cases occur in childhood, i.e. by the age of 15. Varicella is commonly regarded as a mild disease. Serious consequences are rare The incidence of severe complications increases with age Reactivation of dormant VZV results in herpes zoster (shingles) Immune response to VZV can be boosted via two mecanisms : Sub-clinical reactivation of the virus ( endogeneous boosting ) Exposure to infectious individuals ( exogeneous boosting )

8 Vaccination strategies (1) Mass childhood vaccination ± catch-up May lead to an increase in adult cases Vaccination occurs in the youngest ages As the risk of infection decreases, the average age at which exposure to varicella occurs increases Waning vaccine-induced immunity may lead to a pool of susceptible older individuals Significant number of breakthrough cases may occur The lost of exogeneous boosting may lead to an increase on the incidence of zoster

9 Age specific incidence of zoster for adults with and without children in their household Zoster incidence rate per 100,000 persons-yeras 1, No children Children < 16 years RI 0.75 ( ) See also case-control study by Thomas (2002) Age group Brisson et al. Vaccine 2002

10 Vaccination strategies (2) Targeted vaccinations: susceptible adolescents and adults, health care workers, immunocompromised individuals More recently, combined vaccinations Varicella vaccination (in childhood) and zoster vaccination (at older age)

11 What do we learn from US? Routine childhood varicella vaccination program has been introduced in US since 1995 (single injection) The recommendations for vaccination were strengthened in 1999 (varicella vaccination required before entry of a child into day care or school) In June 2006, the vaccination schedule has changed (2 doses)

12 Monthly Varicella Cases and Vaccine Coverage Antelope Valley, CA, Cases Vaccine coverage Jumaan- CDC 2005

13 Reduction in varicella incidence Age Antelope Valley Travis County West Philadelphia < Total Seward JAMA 2002

14 Annual Hospitalization rates per 100,000 population Varicella Active Surveillance Project, Hospitalization Rate/100, ,5 3 2,5 2 1,5 1 0, Jumaan- CDC 2005

15 Number of Varicella Deaths, Number of deaths Average 107 deaths/year Year Jumaan- CDC 2005

16 Adjusted annual rates of breakthrough varicella for children vaccinated between 12 mo-12 yrs: Antelope Valley, Gershon & Arvin, ACIP 2006

17 Distribution of varicella cases by age: Antelope valley, CA, 1995 and 2004 D.Guris ACIP 2006

18 Annual incidence of herpes zoster by age CV 23% CV 48% CV 89% Incidence : 2,77/ 1000 in 1999, 5,25/ 1000 in 2003 (+90%) Yih WK BMC Public Health 2005

19 Epidemiological model (1) Realistic Age Structured Dynamic Model Not vaccinated Protected by maternal antibodies Not vaccinated Susceptibles Not vaccinated Not vaccinated Not vaccinated Latent Infectious Immune infected M(a,t) δ S(a,t) λ (a,t) E(a,t) σ I(a,t) ν R(a,t) f(a,t) g(a,t) kλ (a,t) V(a,t) τ Sv(a,t) bλ (a,t) Ev(a,t) σ Iv(a,t) ν Rv(a,t) Vaccinated Protected by the vaccine Vaccinated Vaccinated Vaccinated Vaccinated Partly Latent Infectious Immune Susceptibles infected t)

20 Epidemiological model (2) Zos Immune α Zoster ρ(a) Sus Boosting zλ(a,t) δ λ(a,t) σ α Susceptible Latent Infectious Immune Τ V Protected W kλ(a,t) bλ σ α V Suscept V Latent V Infectious V Immune

21 Published economic evaluations Since 2000 Country Herpes zoster Waining Herd immunity model Scuffham et al. (2000) Australia No No No Static Brisson et al (2002) Canada Yes Yes Yes Dynamic Getsios et al (2002) Canada No Yes No Static Banz et al. (2003) Germany No Yes Yes Dynamic Hsu et al. (2003) Taiwan No No No Static Coudeville et al. (2004) Italy No Yes Yes Dynamic Coudeville et al. (2005) France & Germany No Yes Yes Dynamic Lenne et al. (2006) Spain No Yes Yes Dynamic Hammerschmidt et al (2007) Germany No Yes Yes Dynamic

22 Models predictions (1) Incidence of varicella would rapidly decline following the implemantation of vaccination After this honeymoon period, a rise in incidence would occur The post-vaccination equilibrium is always lower than the pre-vaccination level The age at infection is predicted to increase but the large reduction in incidence in children is likely to outweight any increase in incidence of chickenpox in adults

23 Impact on incidence: Base case scenario, VC = 90% Varicella cases (per 1,000,000 person-years) 14,000 12,000 10,000 8,000 6,000 4,000 2,000 - Infant Catch-up Adolescent Year after start of vaccination Brisson et al. J Med Virol 2003

24 Impact on average age at infection: Infant vaccination, VC = 90% 14,000 Cases per 1,000,000 personyears 12,000 10,000 8,000 6,000 4,000 2, Year after start of vaccination to 44 5 to 18 0 to 4 Brisson et al. J Med Virol 2003

25 Impact on hospitalizations: Base case scenario, VC = 90% 350 Variclla inpatient days (per 1,000,000 person-years) Infant Catch-up Adolescent Year after start of vaccination Brisson et al. J Med Virol 2003

26 Impact on herpes zoster incidence Herpes zoster cases (per 1,000,000 persons-years) 5,000 4,500 4,000 3,500 3,000 2,500 2,000 1,500 1, Infant, immunity 20 years 95% IC, immunity 7-41 years Adolescent, immunity 20 years Years after start of vacciantion Brisson et al. J Med Virol 2003

27 Models predictions (2) Impact of varicella vaccination on zoster incidence? No Yes Varicella vaccination is less costeffective from the health care payers perspective than from the societal perspective (time / leisure costs) Increases in zoster after routine vaccination of infants would render immunization highly cost-ineffective Routine vaccination of preteens is the preferred strategy under the health care payers perspective

28 Models predictions (4) Persp Infant Catch-up Preteen Infant (BV) Infant (zoster) HC Society ICER ($ / LYG) Brisson et al. J Med Virol 2003

29 Models predictions (5) Infant vaccination (B/C ratio) Country Societal Health care payers Scuffham et al. (2000) Australia Brisson & Edmunds (2002)* Canada Getsios et al. (2002) Canada Hsu et al. (2003) Taiwan Banz et al. (2003) Germany Coudeville at al. (2004) Italy Without taking account zoster

30 Sensitivity analysis Epidemiological factors: Vaccine efficacy Vaccination coverage Economic factors Vaccine price Value of time/leisure lost (indirect costs) Discount rate for health consequences and time horizon Predictive value of anamnestic screening (for targeted vaccinations)

31 Indirect costs (1) Should indirect costs be included in costeffectiveness analysis? If yes, how to measure productivity lost? Human-capital method: Reference to welfare economics (wage rate = opportunity cost of time) Easy to implement Potential productivity costs Friction-costs method: More realistic estimates of productivity costs More data need to be collected The two methods lead to similar results when the period of absenteism is short (as in chickenpox case)

32 Indirect costs (2) Indirect costs = driver for varicella vaccine demand? Few WTP studies in the field Revealed preferences versus real demand (eg Australia) Impacts on vaccination coverage (if CV is not high enough => increase in average age at infection)

33 Comparison of studies Transferability versus generalisability Differences between studies: Hypothesis on key parameters (vaccine efficacy / waining / vaccination coverage) Epidemiological factors Organization of health care systems (impact on costs)

34 0.4 FORCE OF INFECTION ACCORDING AGE Force of infection <1 year 1-2 years 3-4 years 5-9 years years years >= 20 years GERMANY FRANCE

35 Treatment costs per varicella case according to age ( 2002) 120 FRANCE 1.9 GERMANY Additional examinations Hospitalisations Pharmaceutical prescriptions Medical consultations years >= 18 years 1-12 years > 12 years Varicella treatment costs Range % Supported by HS 63% 70% 100% 100%

36 Indirect costs per varicella case according to age ( 2002) 1000 FRANCE GERMANY indirect costs Societal perspective , ,0 Health system perspective ,0 1,5 90,2 105, years >= 18 years 1-12 years > 12 years Indirect costs 97,0 933,4 105,0 795,0 Range % Supported by HS 1,5% 9,7% 70% 0% Coudeville et al. (2004)

37 Average annual costs related to varicella (millions of 2002 ) according to coverage rate Germany Payer s perspective: ,7 42,2 0 No vaccination 5,4 4,7 35,3 High coverage rate (90%) Sick leaves costs Treatment costs Vaccination costs 13,4 12,1 27 Medium coverage (70%) 28,3 24,4 16,8 Low coverage (45%) ,9 1,4 2, France- Payer s perspective: 23,8 4,1 8,2 18 5,1 15,4 11 No vaccination High coverage rate (90%) Medium coverage (70%) Low coverage (45%)

38 Average annual costs related to varicella (millions of 2002 ) according to coverage rate Germany Societal perspective: Sick leaves costs Treatm ent costs Vaccination costs 16 0 France Societal perspective: , ,2 0 No vaccination 15,2 4,7 35,3 High coverage rate (90%) 41,3 12,1 27 Medium coverage (70%) 70,4 24,4 16,8 Low coverage (45%) ,8 39,1 0 No vaccination 19,3 4,2 36,7 High coverage rate (90%) 56,7 12,6 27,8 Medium coverage (70%) 84,6 24,1 16,9 Low coverage (45%)

39 Epidemiological model Varicella: 1 Year old Zoster: 60 years old α ρ(a) Zos Immune Zoster Sus Boosting zλ(a,t) δ λ(a,t) σ α Susceptible Latent Infectious Immune δv e zv VZ Protected Τ V Protected W bλ σ α V Suscept V Latent V Infectious V Immune kλ(a,t) e zv VZ Protected Melegaro A, Jan Van Hoek A, Gay N, Edmunds J (2007)

40 1200 Varicella incidence per 100, Brisson mixing Polymod mixing Combined strategies (2) Mixing patterns Zoster vaccination Years after start of vaccination Preliminary results Melegaro A, Jan Van Hoek A, Gay N, Edmunds J (2007) Zoster increase in incidence Brisson mixing Polymod mixing - infant Polymod mixing - infant + HZ Years after start of vaccination