From the modellers’ point of view - How to communicate with the policy community? - Kari Auranen and Tuija Leino
From the modellers’ point of view -How to communicate with the policy community? Kari Auranen and Tuija Leino Department of Vaccination and Immune Protection National Institute for Health and Welfare, Finland
Outline Infectious disease modelling at NPHI/Finland Three recent examples Varicella vaccination HPV vaccination Pneumococcal conjugate vaccination Communicating modelling results to decision makers Main question(s) for modelling Some results and how they were addressed in decision making Lessons learned
Criteria for the introduction of a new vaccineto the national programme in Finland  There is a considerable disease burden which can be prevented by vaccination  The vaccine is safe on the individual level  Vaccination is safe on the population level  Vaccination is cost-effective NACV
Varicella zoster virus Chickenpox Childhood disease (90% cases in children < 10y) Virus remains latent, can activate as herpes zoster Herpes zoster Disease of the elderly Encounters with varicella virus may sustain immunity against herpes zoster Vaccination protects from chickenpox and (subsequent) herpes zoster Under high coverage of vaccination, circulation of the virus ceases The adult population may become susceptible to herpes zoster because lack of boosting
Introducing varicella vaccination? Vaccination against varicella zoster virus has not been part of the national program in Finland Vaccine-specific expert group (2006-2008) The main question for dynamic modelling What would be the effect of varicella vaccination on zoster incidence?
Three types of data Age-specific incidence of varicella infection At what age do individuals first encounter the virus? A contact survey on the social mixing pattern: ”who meets with whom” From whom do they acquire the virus? Age-specific incidence of herpes zoster At what age do individuals get the disease?
Incidence of varicellaProportion with varicella virus At what age do individuals first encounter the virus? Age (years) Davidkin et al.
Pattern of transmission Finland 70+ 65-69 60-64 Age of the contact 55-59 50-54 From whom 45-49 Age of contact 40-44 35-39 30-34 do individuals 25-29 acquire 20-24 15-19 the virus? 10-14 05-09 00-04 70+ 00-04 05-09 10-14 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 Age of participant Age of participant 0.00-0.31 0.31-0.63 0.63-0.94 0.94-1.25 1.25-1.56 Mossong et al, 2008 1.56-1.88 1.88-2.19 2.19-2.50
Incidence of zoster At what age do individuals get zoster? Karhunen et al., 2009
A transmission model + zoster Susceptible, (a,t) Infected, latent Infectious, VZV positive S(a,t) phase L(a,t) I(a,t) with one previous exposure (1-)v1(a,t) (a,t) Herpes zoster, h(a,d) H(a,t) VZV positive with two Vaccinated v2(a,t) previous exposures protected, R(v)(a,t) Removed, (a,t) R(a,t) VZV positive with three v1(a,t) previous v2(a,t) exposures etc. Vaccinated Vaccinated Vaccinated susceptible, latent, L(v)(a,t) infectious, S(v)(a,t) (a,t) I(v)(a,t)
Effects of vaccination Varicella transmission stops in few years Incidence of zoster increases, in excess to that due to aging population The extent of the excess increase depends on the model assumptions 30-85% excess cases in the next 50 years
Excess cases under 2 scenarios Zoster immunity depends solely on exposure to the virus Zoster immunity depends on exposure and aging Karhunen et al., 2009
What happened next… Cost-effective analysis Varicella vaccination was deemed cost effective, even under the worst case scenario for zoster Vaccine-specific Expert Group (2008) Recommendation, with ”consideration of potential increase in zoster incidence” and a reference to potential use of HZ vaccine not necessarily safe for the (elderly) population National Advisory Committee (2009) No agreement: the decision was put on hold
Lessons learned Worst-case scenarios should perhaps not present the absolutely worst outcomes The modelling group intended to maximise certainty However, the worst-case may have been taken as a likely outcome There was actually a wish to remove any reference into zoster risk Efficient communication of one’s own results possible as long as one can participate in boards Commitment to the implications of one’s own modelling results is inevitable (and appropriate)
Human papillomavirus (HPV) Common asymptomatic and usually transient infection Up to 30% of young adults are infected at any time Infection may become persistent and progress to cancer Cervical cancer (5.7/100,000 in Finland) The incidence of disease is greatly modulated by the very effective screening program in Finland Vaccination protects against primary HPV infection and (subsequent) disease The impact of screening is intertwined with that of vaccination
Introducing HPV vaccination? The aims for dynamic modelling To disentangle the underlying disease process from that affected by the current screening program To optimise the screening program To consider the optimal introduction policy for HPV vaccination The model was constructed in two parts (1) Transmission of the virus (2) Progression of infection to disease
Elements of modelling HPV TREATMENT AND MANAGEMENT CIN0Outcome CIN1 CIN2 CIN3 CaScreening testing testing testing testing testing/ sypmtomsRate of CIN0 CIN1 CIN2 CIN3 Cancerinfection Clear Clear Clear Clear Vänskä, Salo et al. 2012
The fate of HPV vaccination(and screening)? Vaccine-specific Expert Group and the National Advisory Committee recommended introducing the vaccine The screening experts presented strong criticism against the underlying analysis Strong reliance on the current screening policy The perception of the burden of disease was based on the apparent incidence of disease Opportunistic screening falls out of the sight of the systematic screening program The final decision was put on hold, primarily for financial reasons
Lessons learned Communicating modelling results to a group of outside stakeholders in their own substance matter area is extremely difficult Without success in engaging the group from the first beginning Without the incentive originating from that group Without expertise and tradition in the methods of modelling in that group If the model-based analysis is not totally transparent, including its relation to epidemiological data Criticism is considered more acceptable Criticism more likely misses the point The justification of criticism is impossible to assess by the third parties
Pneumococcus (Streptococcus pneumoniae) Causes different forms of disease Mild infections of the respiratory tract Pneumonia Meningitis Is usually carried asymptomatically New vaccines protect against disease due 10 or 13 types out of the >90 types Vaccination affects asymptomatic carriage as well This may lead to increase in carriage and disease caused by the non-vaccine types
Setting the tender criteria Currently, the 10-valent vaccine is in the national immunisation program A new tender process was to be prepared to choose between the 10- and 13-valent vaccines Price and quality (i.e. number of types) Questions for modelling: Considering the expected greater health benefits from the 13-valent vaccine, how much more are we prepared to pay for that? What is the expected difference in the health outcomes (incidence of disease)?
Differences in disease incidence Salo et al., 2012
What happened next? The Ministry accepted the tender criteria, based on the modelling results This was based on a simplified model of vaccine efficacy against individual types This fact was overlooked by the board, probably due to too much preoccupance to other assumptions The tender criteria have already been criticised by the other vaccine provider Criticism of the models not being truthful enough!
Lessons learned Although the models incorporate a large number of assumptions, only some of them usually catch attention in expert panels The communication needs to be based on few key assumptions It is not worthwhile to present solutions to problems, which the expert panel does not know to appreciate in the first hand Implications of the decisions need to be made explicit
Concluding remarks Communication for public health decision makers and science audiences differ More emphasis on ”certainty” in public health Peer-reviewed publications from the group give more support and credibility The modeller’s view may sometimes be (appropriately) naive We live in the world of models and methods!
Acknowledgements Heini Salo Simopekka Vänskä Markku Nurhonen Markku Karhunen