- Email: [email protected]
Modeling pandemic preparedness scenarios: health economic implications of enhanced pandemic vaccine supply
Virus Research 103 (2004) 9–15
Modeling pandemic preparedness scenarios: health economic implications of enhanced pandemic vaccine supply Jeroen K. Medema a,∗ , York F. Zoellner b , James Ryan c , Abraham M. Palache a b
a Business Group Influenza, Solvay Pharmaceuticals BV, PO Box 900, Weesp 1380 DA, The Netherlands Department of Health Economics, Solvay Pharmaceuticals GmbH, PO Box 220, Hannover 30002, Germany c Mapi Values, The Adelphi Mill, Bollington, Macclesfield, Cheshire SK105JB, UK
Available online 9 April 2004
Abstract Influenza pandemic planning is a complex, multifactorial process, which involves public health authorities, regulatory authorities, academia and industry. It is further complicated by the unpredictability of the time of emergence and severity of the next pandemic and the effectiveness of influenza epidemic interventions. The complexity and uncertainties surrounding pandemic preparedness have so far kept the various stakeholders from joining forces and tackling the problem from its roots. We developed a mathematical model, which shows the tangible consequences of conceptual plans by linking possible pandemic scenarios to health economic outcomes of possible intervention strategies. This model helps to structure the discussion on pandemic preparedness and facilitates the translation of pandemic planning concepts to concrete plans. The case study for which the model has been used shows the current level of global pandemic preparedness in an assumed pandemic scenario, the health economic implications of enhanced pandemic vaccine supply and the importance of cell culture-based influenza vaccine manufacturing technologies as a tool for pandemic control. © 2004 Published by Elsevier B.V.
1. Introduction Influenza is an acute respiratory disease, which can often lead to serious and life-threatening complications in several populations, like the elderly and chronically ill. It occurs in annual epidemics in regions of temperate climates with illness attack rates of 10–20%, leading to an average of 114,000 hospitalizations and 20,000 excess deaths in the United States alone (Strikas et al., 2002). In addition to these annual epidemics, type A influenza viruses have caused pandemics, i.e. sudden global epidemics in all age groups with higher attack and mortality rates. These pandemics are caused by a newly emerging virus subtype in the human population, resulting from reassortment of gene segments between influenza viruses with different host susceptibility (e.g. human, swine and avian strains) or by direct transmission of non-human virus subtypes to humans. The last century has seen three influenza pandemics, the “Spanish Flu” in 1918–1919, the “Asian Flu” in 1957 and the “Hong Kong Flu” in 1968. The first, one of the most se∗
Corresponding author. Tel.: +31-294-477000; fax: +31-294-431164. E-mail address: [email protected] (J.K. Medema).
0168-1702/$ – see front matter © 2004 Published by Elsevier B.V. doi:10.1016/j.virusres.2004.02.006
vere events in human history in terms of cases and deaths, was responsible for an estimated 2 billion cases and 20–50 million deaths worldwide (Potter, 1998; Davies, 2000). The latter two were less severe; the 1957 pandemic led to over 1 million deaths worldwide and 70–80,000 in the United States, whereas the “Hong Kong Flu” was reported as relatively mild with peak mortality rates similar to 1957 (Potter, 1998; Strikas et al., 2002). As the influenza virus was only discovered in 1933, influenza diagnosis was poor before that date; hence the corresponding clinical cases have neither been identified nor documented as such before the 20th century. Analysis of historical documentation, however, clearly indicates that influenza pandemics have occurred at irregular intervals over many centuries and there are no reasons to doubt they will occur again. The increased awareness of the societal burden of annual influenza epidemics since the mid-1990s led to an increasing awareness of the pandemic threat we are facing. Furthermore, the emergence of two new influenza viruses—subtype H5N1 in 1997 and subtype H9N2 in 1999—in humans in the Hong Kong area, which seem to be transmitted directly from avian species, made this threat real and imminent. The events following 11 September 2001 have further contributed by
10
J.K. Medema et al. / Virus Research 103 (2004) 9–15
drawing the general public’s attention to the hazard of biological threats. The increased awareness on influenza pandemics has led to discussions by public health authorities, regulatory authorities, academia and industry on what our society can and should do to prepare for the next pandemic. This has resulted amongst others in the founding of the Influenza Vaccine Supply (IVS) Taskforce, a new pandemic discussion and collaboration platform in which all major influenza vaccine manufacturers are represented. Increased awareness has also resulted in governmental pandemic preparedness plans on national and regional levels, such as the “Influenza Pandemic Preparedness Action Plan for the United States” and “A pandemic influenza planning guide for state and local officials” by the Centers of Disease Control and Prevention (Strikas et al., 2002; Patriarca et al., 1999) and the “Canadian contingency plan for pandemic influenza” (Health Canada, 2000). Several attempts are undertaken to bring pandemic planning to multinational or even global level, such as the November 2001 conference on “Preparedness planning in the EU: Influenza and other health threats” and the WHO influenza pandemic preparedness plan (World Health Organization, 1999). Although some regions, e.g. the Canadian province of Ontario, are quite well advanced in pandemic planning and implementation, most of both national and international plans are not completed, let alone realized. Pandemic planning is a multifactorial process with a consequent high complexity. It is further complicated by the unpredictability of the next pandemic; its time of emergence, spread and severity cannot be foreseen and the efficacy of interventions available for annual influenza epidemics cannot be extrapolated per se to a pandemic situation. Both the complexity and the uncertainties surrounding pandemic preparedness have so far kept the various stakeholders from joining forces and tackling the problem from its roots. The complexity of preparedness planning must be simplified by dividing it into several “sub-projects”, which should be addressed separately to bring pandemic planning from abstract concepts to the desired concreteness. We developed a mathematical model, which shows the tangible consequences of conceptual plans by linking possible pandemic scenarios to health economic outcomes of possible intervention strategies. This model helps to structure the discussion on pandemic preparedness by calculating costs and benefits, gauges the public health benefits of optimized preparedness, and facilitates the translation of pandemic planning concepts to concrete plans.
2. Materials and methods A computer-based simulation model has been developed by MAPI Values (Macclesfield, United Kingdom) and Solvay Pharmaceuticals (Weesp, The Netherlands and Hannover, Germany), which allows for epidemiological, cost and efficacy inputs as well as manufacturing con-
straints. The model combines a vaccine production model with an adapted and revised cost-effectiveness model of influenza intervention strategies, based on that developed by Scuffham and West (2002). 2.1. Inputs for pandemic scenarios The model inputs for the pandemic scenarios determine the characteristics of the pandemic to which the several intervention strategies can be directed; these input parameters are the attack and mortality rate of that particular pandemic virus, the time of emergence of the pandemic virus and the time for the pandemic to spread and end. 2.2. Inputs for intervention scenarios The model inputs for intervention scenarios determine the availability of possible interventions, e.g. pandemic vaccine and antivirals, such as amantadine or the new neuraminidase inhibitors. These input parameters include specific vaccine production inputs, such as time to produce a suitable pandemic virus seed, time to produce a batch of vaccine, availability of eggs for egg-based vaccine production and available vaccine manufacturing capacity. They do not include specific antiviral production inputs, as it is assumed that if antivirals are selected as intervention strategy, they will be stockpiled and readily available. Intervention scenario inputs also include parameters, such as pandemic vaccine potency and dosing regime. It is assumed that all manufactured vaccines are used. 2.3. Inputs for intervention effectiveness Inputs for intervention effectiveness include parameters such as reduction of influenza-like illness (ILI) cases, reduction in primary care physician (PCP) visits for ILI symptoms, reduction of hospitalizations due to influenza and pneumonia, other respiratory illnesses and congestive heart failure. 2.4. Health economic inputs The model allows direct costs, from the perspective of the health care system, to be put in. Parameters for direct costs include all medical costs of the burden of the pandemic and for the intervention strategy selected. The medical costs for the burden of the pandemic include input parameters like costs of PCP consultations, costs of hospitalization due to influenza and pneumonia, other respiratory illnesses and to congestive heart failure. By combining average costs with resource use (PCP consultations, hospitalizations, etc.), the total medical costs for that particular pandemic can be calculated. The medical costs for the intervention strategy include the vaccine and vaccination cost as well as the costs of antiviral prophylaxis or therapy. The model allows for a series of secondary input parameters, such as the probability of having an antiviral prescribed
J.K. Medema et al. / Virus Research 103 (2004) 9–15
during a PCP visit with ILI symptoms, proportion of PCP visits undertaken at home and days off work per case. 2.5. Approach One way of giving structure to conceptual pandemic preparedness plans is to evaluate these plans from a health economic perspective. Therefore the developed model is used to calculate the costs and benefits of certain intervention scenarios for different pandemic scenarios. The strength of the model is its ability to visualize different pandemic scenarios and different preparedness strategies, which is reinforced by assuming the pandemic threat is imminent, rather than taking place at an unknown point in time decades from now. Hence we assumed the next pandemic strain emerges 1 January 2004, in order to show the current level of pandemic preparedness and how the model can visualize this. We assumed that the pandemic scenario caused by the “2004 pandemic” virus is based on the 1918–1919 pandemic with influenza attack rates of 20–50% depending on region and a case fatality rate of 1.5% in two pandemic waves (Potter, 1998). For the “2004 pandemic” an average 1918–1919 attack rate of 35% in a single wave with a case fatality rate of 1.87% (Scuffham and West, 2002) are put into the model as pandemic scenario parameters. Health economic parameters put into the model are based on average data obtained in the elderly in the United Kingdom in recent annual influenza epidemics (Scuffham and West, 2002). For the “2004 pandemic” case study it is assumed that health services utilization during the influenza pandemic are comparable to those during annual epidemics; increased absenteeism of health care workers and societal disruption by the pandemic are not taken into account. It is acknowledged that health care costs and health economical consequences of influenza in the United Kingdom cannot be translated to other countries, especially not to developing countries. Therefore, the latter are left out of this case study and the results that come from this study only visualize the level of preparedness of the developed countries. The total population of the developed countries is estimated at 1 billion. The key input parameters for the model, including vaccination efficacy, are listed in Table 1.
3. Results The outputs of the model that show the burden of the assumed “2004 pandemic” on developed countries are listed in Table 2 under the no intervention scenario. These data show that the no intervention scenario leads to 350 million cases and 22.2, 48.2 and 6.55 million influenza-related PCP consultations, hospitalizations and excess deaths, respectively. Total direct medical care costs add up to 166.6 billion, with hospitalizations costs of 165.2 billion as the major contributor.
11
Table 1 Key input parameters used within the model Vaccine production Month of pandemic strain emerging Month egg seed released by WHO Month last vaccine required in market Number of weeks to produce a vaccine Weeks cell seed released before egg seed Size of (trivalent) vaccine unit (g) Number of shots required per person Dose per shot (g) Annual manufacturing capacity
January 2004a,b April 2004a July 2004a,b 5a,b 9b 45a,b 2a,b 7.5a,b 230,000,000a,b
Demographics Overall population Target population Influenza attack rate (%) Years per life lost
1,000,000,000a,b 1,000,000,000a,b 35a,b 5c
Event probabilities PCP consultations (for influenza) Hospitalization rates due to: Influenza + pneumonia Other respiratory disease Congestive heart failure Mortality rate Vaccine effectiveness Vaccine efficacy (%) Reduction in GP visits (%) Reduction in hospitalizations due to: Influenza + pneumonia (%) Other respiratory disease (%) Congestive heart failure (%) Reduction in mortality (%) Unit costs and economic parameters PCP consultation ( ) Influenza vaccine price ( ) Influenza vaccine service and administration ( ) Discount rate Hospitalization for: Influenza + pneumonia ( ) Other respiratory disease ( ) Congestive heart failure ( ) Average hospital cost per bed day ( )
6.34d 2.52d 9.34d 1.9d 1.87d 53e 46f 39f 32f 27f 50f 28.6c 10a,b 3.1c 5c 3585c 3362c 3556c 261c
a
Authors’ assumption for egg-based vaccine intervention strategies. Authors’ assumption for cell culture-based vaccine intervention strategies. c Scuffham and West (2002). d Derived from Scuffham and West (2002). e Govaert et al. (1994). f Nichol et al. (1998). b
To visualize the current level of pandemic preparedness, the intervention strategies that can be launched to decrease the burden of the “2004 pandemic” are based on existing egg-based influenza vaccine technologies. Currently, the estimated Northern Hemisphere influenza vaccine usage is 230 million trivalent doses, which translates to a global influenza vaccine manufacturing capacity of 22 million monovalent doses (i.e. 15 g of pandemic strain antigen) per week. Two main starting materials are needed to manufacture the pandemic vaccine with the current technology: a sufficient amount of embryonated hen’s eggs bred under controlled
12
J.K. Medema et al. / Virus Research 103 (2004) 9–15
Table 2 Burden of different intervention scenarios in the assumed pandemic
Vaccination coverage (%) Number vaccinated Number of cases (influenza) PCP visits for treatment (influenza) Hospitalizations Excess deaths Discounted years per life lost PCP visit costs—vaccinations ( ) PCP visits costs—treatment ( ) Hospitalization costs ( ) Strategy costs ( ) Medical care costs ( ) Total direct costs ( )
No intervention
Intervention scenario 1 Egg-based vaccine manufacture
Intervention scenario 2 Cell culture-based vaccine manufacture
0 0 350,000,000 22,190,000 48,160,000 6,545,000 25,640,894 0 1,007,181,910 165,170,880,000 0 166,647,295,200 166,647,295,200
17 170,000,000 320,250,000 20,454,742 45,491,663 5,988,675 23,461,418 2,227,000,000 928,420,285 156,010,312,598 2,227,000,000 157,371,272,130 159,598,272,130
37 370,000,000 285,250,000 18,413,262 42,352,443 5,334,175 20,897,328 4,847,000,000 835,759,549 145,233,174,478 4,847,000,000 146,458,303,811 151,305,303,811
conditions and a seed of the virus strain to which the vaccine is directed. Preparation of the latter is performed by WHO collaborating centers and takes 1.5–2.5 months for annual epidemic strains. For the assumed “2004 pandemic” seed preparation time is unknown and unpredictable; we assumed a seed preparation time of 3 months, which means that pandemic vaccine manufacture can start 1 April 2004. This is within the regular annual influenza vaccine manufacturing season, which means that the eggs are readily available for pandemic vaccine manufacture. We assumed that the applied “2004 pandemic” vaccine will be 7.5 g monovalent given in two doses and should be administered within 6 months after emergence of the “2004 pandemic” virus. The results of the assumed intervention strategy on the “2004 pandemic” are listed in Table 2 as intervention scenario 1. These data show that the current level of pandemic preparedness in the assumed egg-derived scenario leads to vaccination of 17% of the population, which avoids 29.8 million influenza cases, 1.74 million PCP visits, 2.67 million hospitalizations and 556,000 deaths compared to a scenario of no vaccination. The intervention strategy costs add up to 2.2 billion, but as this strategy leads to a reduction of 9.3 billion medical care costs, the saving on direct costs of the “2004 pandemic” is 7.1 billion. The question for pandemic planners is if the level of pandemic preparedness can be improved and which intervention strategies should be selected to achieve this. As vaccination is the most cost-effective intervention, an improved intervention strategy from a health economic perspective would be to produce more pandemic vaccine in the available time frame. This can be achieved by increasing the time period for vaccine manufacture, by increasing the vaccine manufacturing capacity or preferably by a combination of the two. An earlier availability of a suitable seed virus would increase the time frame for pandemic vaccine manufacture and hence improve the intervention strategy. The new cell culture-based influenza vaccine manufacturing technology likely facilitates seed preparation, as mammalian cell culture may require less adaptation of viruses and allows con-
tainment of more pathogenic pandemic viruses without prior decrease of their virulence (see Section 4). To visualize this we calculated the level of pandemic preparedness with an intervention scenario of cell culture-based vaccine manufacturing, assuming that preparation of a cell culture-based seed virus takes 1 month and assuming that the current global weekly manufacturing capacity of 22 million monovalent doses is cell culture-based in stead of egg-based. The results of this assumed intervention strategy on the “2004 pandemic” are listed in Table 2 as intervention scenario 2. These results show that, compared to no intervention scenario, the cell culture-based intervention strategy avoids 75 million influenza cases, 3.78 million PCP consultations for influenza treatment and, respectively, 5.81 million and 1.21 million influenza-related hospitalizations and excess deaths. Compared to the assumed egg-based vaccine intervention with 17% vaccine coverage, the cell culture-based intervention strategy leads to vaccination of 37% of the population, avoiding an additional 35 million influenza cases, 2.04 million PCP consultations for influenza treatment, 3.14 million influenza-related hospitalizations and 654,500 excess deaths. The cell culture-based intervention strategy costs add up to 4.9 billion, but as this strategy leads to a reduction of 20.2 billion medical care costs, the saving on direct costs of the “2004 pandemic” is 15.3 billion, an additional 8.3 billion compared to the egg-based intervention scenario. The years per life lost gained are 2.56 million, whereas the cost per life year gained is 3198. The cost per case, hospitalization and death averted are, respectively, 234, 2612 and 12,530.
4. Discussion The “2004 pandemic” case study is based on numerous assumptions. Each can be challenged and, indeed, such uncertain variables are discussed by pandemic planners to challenge preparedness concepts. However, the assumed parameters are only inputs to the presented model and not part of
J.K. Medema et al. / Virus Research 103 (2004) 9–15
the model itself. Therefore, questioning these assumptions is not questioning the model itself. The model is only a tool to visualize the consequences and effectiveness of a number of planning concepts, to which it lends structure and logic. Moreover, a sensitivity analysis of the model can be used to set priorities within pandemic planning discussions; the more sensitive the model is to a given assumption, the higher the effect of that particular input on the level of pandemic preparedness. A sensitivity analysis of the “2004 pandemic” case study shows that the model is most sensitive to the pandemic scenario in terms of attack and mortality rates, as well as to the intervention scenarios in terms of effectiveness and available amounts of pandemic vaccine. It is less sensitive to medical care costs and relatively insensitive to pandemic vaccine price. The total direct medical costs of 166.6 billion are remarkably close to the 166.5 billion calculated by the CDC for a pandemic with 35% attack rate (Meltzer et al., 1999). However, these are for the United States alone with 89 million cases, 734,000 hospitalizations and 207,000 excess deaths on a smaller population than in the model presented here. Although the CDC model is similar, the inputs are based on US data, explaining the higher costs. 4.1. Pandemic scenario The assumptions put into the model to derive the presented “2004 pandemic” scenario have been based on the 1918–1919 pandemic in terms of spread and severity. The first cases were reported in March 1918, leading to a first wave that was not severe. It took about 5–6 months before the pandemic evolved into a second wave with the particular characteristics of high attack and mortality rates, in the age groups at risk for epidemic influenza as well as other age groups, like children and young adults (Potter, 1998). This pandemic shows that everybody can be at risk for pandemic influenza and means that at that time intervention in all age groups at risk should have taken place in these 6 months between emergence and actuality of the pandemic. However, both knowledge and surveillance of circulating influenza viruses have been well established over the last 50 years, enabling an earlier identification of potential pandemic viruses and consequently an earlier onset of intervention programs compared to 1918. Furthermore, improved influenza diagnosis and the availability of intervention enable us to significantly reduce pandemic virus spread. On the other hand, the increased globalization and international travel in current times are likely to speed up spread of pandemic viruses to a large extent. As these phenomena counteract, we assumed for the “2004 pandemic” the same time frame between emergence and spread of the pandemic virus as in 1918. The severity of a pandemic is reflected in the model by attack rate and event probabilities, such as influenza-related PCP consultations, hospitalizations and case fatality rate. For the “2004 pandemic” we assumed a higher attack rate
13
to reflect a higher severity compared to annual epidemics. All event probabilities for the “2004 pandemic” are based on annual epidemic UK event probabilities for the elderly, which are higher than for other age groups. However, the higher attack rate implies that during the “2004 pandemic” other age groups are affected as well and extrapolation of high event probabilities in the elderly to these age groups also reflects a more severe event. 4.2. Egg-based vaccine intervention scenario Health economic studies on annual influenza epidemics have shown that influenza vaccination is by far more cost-effective than prescription of antivirals for prophylaxis or therapy (Scuffham and West, 2002). Because the health economic perspective has been used in the model to visualize the current level of pandemic preparedness, vaccination has been selected as the pandemic intervention strategy in the current case study as well. The case study is limited to the developed world, because the health economic inputs are based on data gathered in a developed country, but pandemics are a global event and the burden in developing countries might be even greater because of lower public health. However, as long as global pandemic preparedness plans have not been realized, the world will face a limited availability of interventions. These will be made available to developed countries first at the cost of the developing world. The intervention scenario input parameters are based on current influenza vaccine manufacturing technologies. We assumed that the applied “2004 pandemic” vaccine will be 7.5 g monovalent given in two doses, as the human population is immunologically na¨ıve to the pandemic strain and will need a booster-dose to elicit a protective immune response. Clinical studies in unprimed individuals indeed indicate that a two dose regime is needed (Nicholson et al., 1979), but that this may be achieved with a lower, possibly adjuvanted dose (Nicholson et al., 2001; Hehme et al., 2002). The next pandemic is likely to be detected by the WHO global influenza surveillance network and, just as for annual epidemic strains, WHO collaboration centers prepare the pandemic virus seed. Seed preparation time for the next pandemic is unpredictable and cannot be based on experience with epidemic strains; although the H5N1-subtype that emerged in Hong Kong in 1997 appeared only a candidate pandemic virus, a suitable seed is still not available 5 years later. In case of a high pandemic threat the efforts that will be put into seed preparation will be accordingly and preparation times for pandemic seeds suitable for egg-based vaccine manufacture in general is expected to be 2–8 months (internal communication J. Woods of WHO Collaboration Center NIBSC, WHO Experts-IVS Taskforce Meeting, 26 September 2002, Geneva). For the case study we assumed 3 months, leaving 3 months for vaccine manufacture, distribution and vaccination before the “2004 pandemic” would evolve and widely spread.
14
J.K. Medema et al. / Virus Research 103 (2004) 9–15
4.3. Cell culture-based vaccine intervention scenario Field strains circulating in the human population usually do not propagate well on eggs and need to be attenuated for egg-based vaccine manufacture. Therefore, both epidemic and pandemic seed preparation for egg-based technology mainly consists of preparation of so-called high growth reassortants by reassortment of field strains with strains that propagate well on eggs. Additionally, pandemic viruses are more virulent than epidemic strains and need to be attenuated to decrease their virulence to safely propagate such viruses in existing egg-based manufacturing facilities, as these are based on open systems. On the contrary, cell culture-based vaccine manufacture is performed in closed bioreactor systems, enabling the desired containment to process more virulent pandemic viruses without prior attenuation and hence less laborious seed preparation. Moreover, cell culture-based influenza vaccine manufacturing technologies make use of mammalian cell lines, such as MDCK, which are more closely related to the human host than the avian egg-based technology. MDCK-grown human influenza field isolates indeed are generally more antigenically homogenous and more alike the original field strain than egg-grown isolates (Zambon, 1998), which is probably caused by host cell selection of variants. Mammalian epithelioid cells like MDCK appear to be the most sensitive cell culture system for human influenza viruses to date (Zambon, 1998), whereas some human influenza isolates need to be adapted to growth in the allantoic cavity (Murphy and Webster, 1996). Therefore, cell culture is generally used for primary isolation of human influenza viruses. The probability that human pandemic viruses readily grow in mammalian cells therefore seems higher than in the more distant avian host, thereby making the preparation of high growth reassortants obsolete and decreasing the time for pandemic seed preparation. This favors the preparation of pandemic seeds by propagation of the pandemic strain directly on mammalian cell culture in stead of via embryonated eggs. However, for annual epidemic vaccine manufacture the egg-passage is a regulatory obligation; omission of this procedure therefore needs to be cleared by authorities. As the time period for vaccine manufacture is mainly determined by the time a suitable virus seed is available and the time the pandemic spreads in the population, a potential quicker release of pandemic seed suitable for cell culture enables the production of more vaccines. The more vaccines, the greater the vaccination rate and the greater the opportunity to benefit from both reduced mortality and reduced total costs. This underlines the importance of cell culture-based influenza vaccine manufacturing as tool for increased pandemic preparedness. By assuming the pandemic virus emerges in January, eggs are readily available for pandemic vaccine manufacture once the seed is available. However, a pandemic virus can emerge all year round, also outside the annual epidemic vaccine manufacturing period. Given that logistics of egg
preparation require ordering about 1 year in advance, availability of sufficient eggs outside the planned period certainly can be questioned. Cell culture-based vaccine manufacture however makes long-term advance planning obsolete, as all starting materials are in stock and readily available whenever required. This advantage however is not illustrated in the assumed “2004 pandemic”, thereby underestimating the advantageous health economics of cell culture-based over egg-based intervention and thus the importance of cell culture-based vaccine manufacturing as tool for increased pandemic preparedness. Although the presented model can lend structure and logic to pandemic preparedness discussions, pandemic planning remains a complex process. The efficacy of interventions, such as vaccines and antivirals, during pandemics needs to be studied and proven beforehand as much as possible in order to make such interventions available in time. Regulatory procedures of such interventions need to be harmonized and adapted to the short-time lines available in pandemic situations and logistical systems need to be set up or streamlined to successfully execute the intervention strategy. Pandemic vaccination might be the most cost-effective approach, but will require the availability of a suitable virus seed and adequate manufacturing facilities to process such a seed. In order to manufacture sufficient amounts for an adequate level of pandemic vaccine, manufacturing capacity needs to be increased. As stated by WHO, policy makers need to keep in mind the several years needed to construct new production facilities and significantly increase production capacity (World Health Organization, 2002). This only has an economical incentive if interpandemic vaccine usage is increased, which—just as pandemic preparedness—is a joint responsibility of the public and private sector.
Acknowledgements The authors like to thank James Piercy of Mapi Values for the health economic support during development of the model and Professor Kristin L. Nichol of the University of Minnesota Medical School for her valuable comments on the manuscript.
References Davies, P., 2000. Catching Cold, 1918s Forgotten Tragedy and the Scientific Hunt for the Virus that Caused It. Penguin Books, Harmondsworth, UK, 306 pp. Govaert, T.M., Sprenger, M.J., Dinant, G.J., Aretz, K., Masurel, N., Knottnerus, J.A., 1994. Immune response to influenza vaccination of elderly people. A randomized double-blind placebo-controlled trial. Vaccine 12, 1185–1189. Health Canada, 2000. Canadian Contingency Plan for Pandemic Influenza—Draft January 5, 2000. The Laboratory Centre for Disease Control, Health Canada, Ottawa.
J.K. Medema et al. / Virus Research 103 (2004) 9–15 Hehme, N., Engelmann, H., Kunzel, W., Neumeier, E., Sanger, R., 2002. Pandemic preparedness: lessons learnt from H2N2 and H9N2 candidate vaccines. Med. Microbiol. Immunol. 191 (3–4), 203–208. Meltzer, M.I., Cox, N.J., Fukuda, K., 1999. The economic impact of pandemic influenza in the United States: priorities for intervention. Emerg. Infect. Dis. 5, 659–671. Murphy, B.R., Webster, R.G., 1996. Orthomyxoviruses. In: Fields, B.N., Knipe, P.M., Howley, P.M. (Eds.), Fields Virology, 3rd ed. Lippincott-Raven, Philadelphia, pp. 1397–1445. Nichol, K.L., Wuorenma, J., Von Sternberg, T., 1998. Benefits of influenza vaccination for low-, intermediate-, and high-risk senior citizens. Arch. Intern. Med. 80, 1769–1776. Nicholson, K.G., Tyrrell, D.A.J., Harrison, P., 1979. Clinical studies of monovalent inactivated whole virus and subunit A/USSR/77 (H1N1) vaccine: serological responses and clinical reactions. J. Biol. Stand. 7, 123–136. Nicholson, K.G., Colegate, A.E., Podda, A., Stephenson, I., Wood, J., Ypma, E., Zambon, M.C., 2001. Safety and antigenicity of non-adjuvanted and MF59-adjuvanted influenza A/Duck/Singapore/97 (H5N3) vaccine: a randomized trial of two potential vaccines against H5N1 influenza. Lancet 357, 1937–1943. Patriarca, P.A., Strikas, R.A., Gensheimer, K.F., Cox, N.J., Fukuda, K., Meltzer, M.I., 1999. A pandemic influenza planning guide for
15
state and local officials. Draft 2.1—National Vaccine Program Office, Centers for Disease Control and Prevention, Atlanta (available at http://www.cdc.gov/od/nvpo/pub.htm). Potter, C.W., 1998. Chronicle of influenza pandemics. In: Nicholson, K.G., Webster, R.G., Hay, A.J. (Eds.), Textbook of Influenza. Blackwell Science, Oxford, pp. 3–18. Scuffham, P.A., West, P.A., 2002. Economic evaluation of strategies for the control and management of influenza in Europe. Vaccine 20, 2562–2578. Strikas, R.A., Wallace, G.S., Myers, M.G., 2002. Influenza pandemic preparedness action plan for the United States: 2002 update. CID 35, 590–596. World Health Organization, 1999. Influenza Pandemic Preparedness Plan: The Role of WHO and Guidelines for National and Regional Planning. WHO/CDS/CSR/EDC/99.1, Geneva (available at http://www. who.int/emc-documents/influenza/whocdscsredc991c.html). World Health Organization, 2002. Draft WHO Guidelines on the Use of Vaccines and Antivirals during Influenza Pandemics. Issued 2–4 October 2002, Geneva (available at http://www.who. int/emc/diseases/flu/whoguidelines.htm). Zambon, M., 1998. Laboratory diagnosis of influenza. In: Nicholson, K.G., Webster, R.G., Hay, A.J. (Eds.), Textbook of Influenza. Blackwell Science, Oxford, pp. 3–18.
Modeling pandemic preparedness scenarios: health economic implications of enhanced pandemic vaccine supply Jeroen K. Medema a,∗ , York F. Zoellner b , James Ryan c , Abraham M. Palache a b
a Business Group Influenza, Solvay Pharmaceuticals BV, PO Box 900, Weesp 1380 DA, The Netherlands Department of Health Economics, Solvay Pharmaceuticals GmbH, PO Box 220, Hannover 30002, Germany c Mapi Values, The Adelphi Mill, Bollington, Macclesfield, Cheshire SK105JB, UK
Available online 9 April 2004
Abstract Influenza pandemic planning is a complex, multifactorial process, which involves public health authorities, regulatory authorities, academia and industry. It is further complicated by the unpredictability of the time of emergence and severity of the next pandemic and the effectiveness of influenza epidemic interventions. The complexity and uncertainties surrounding pandemic preparedness have so far kept the various stakeholders from joining forces and tackling the problem from its roots. We developed a mathematical model, which shows the tangible consequences of conceptual plans by linking possible pandemic scenarios to health economic outcomes of possible intervention strategies. This model helps to structure the discussion on pandemic preparedness and facilitates the translation of pandemic planning concepts to concrete plans. The case study for which the model has been used shows the current level of global pandemic preparedness in an assumed pandemic scenario, the health economic implications of enhanced pandemic vaccine supply and the importance of cell culture-based influenza vaccine manufacturing technologies as a tool for pandemic control. © 2004 Published by Elsevier B.V.
1. Introduction Influenza is an acute respiratory disease, which can often lead to serious and life-threatening complications in several populations, like the elderly and chronically ill. It occurs in annual epidemics in regions of temperate climates with illness attack rates of 10–20%, leading to an average of 114,000 hospitalizations and 20,000 excess deaths in the United States alone (Strikas et al., 2002). In addition to these annual epidemics, type A influenza viruses have caused pandemics, i.e. sudden global epidemics in all age groups with higher attack and mortality rates. These pandemics are caused by a newly emerging virus subtype in the human population, resulting from reassortment of gene segments between influenza viruses with different host susceptibility (e.g. human, swine and avian strains) or by direct transmission of non-human virus subtypes to humans. The last century has seen three influenza pandemics, the “Spanish Flu” in 1918–1919, the “Asian Flu” in 1957 and the “Hong Kong Flu” in 1968. The first, one of the most se∗
Corresponding author. Tel.: +31-294-477000; fax: +31-294-431164. E-mail address: [email protected] (J.K. Medema).
0168-1702/$ – see front matter © 2004 Published by Elsevier B.V. doi:10.1016/j.virusres.2004.02.006
vere events in human history in terms of cases and deaths, was responsible for an estimated 2 billion cases and 20–50 million deaths worldwide (Potter, 1998; Davies, 2000). The latter two were less severe; the 1957 pandemic led to over 1 million deaths worldwide and 70–80,000 in the United States, whereas the “Hong Kong Flu” was reported as relatively mild with peak mortality rates similar to 1957 (Potter, 1998; Strikas et al., 2002). As the influenza virus was only discovered in 1933, influenza diagnosis was poor before that date; hence the corresponding clinical cases have neither been identified nor documented as such before the 20th century. Analysis of historical documentation, however, clearly indicates that influenza pandemics have occurred at irregular intervals over many centuries and there are no reasons to doubt they will occur again. The increased awareness of the societal burden of annual influenza epidemics since the mid-1990s led to an increasing awareness of the pandemic threat we are facing. Furthermore, the emergence of two new influenza viruses—subtype H5N1 in 1997 and subtype H9N2 in 1999—in humans in the Hong Kong area, which seem to be transmitted directly from avian species, made this threat real and imminent. The events following 11 September 2001 have further contributed by
10
J.K. Medema et al. / Virus Research 103 (2004) 9–15
drawing the general public’s attention to the hazard of biological threats. The increased awareness on influenza pandemics has led to discussions by public health authorities, regulatory authorities, academia and industry on what our society can and should do to prepare for the next pandemic. This has resulted amongst others in the founding of the Influenza Vaccine Supply (IVS) Taskforce, a new pandemic discussion and collaboration platform in which all major influenza vaccine manufacturers are represented. Increased awareness has also resulted in governmental pandemic preparedness plans on national and regional levels, such as the “Influenza Pandemic Preparedness Action Plan for the United States” and “A pandemic influenza planning guide for state and local officials” by the Centers of Disease Control and Prevention (Strikas et al., 2002; Patriarca et al., 1999) and the “Canadian contingency plan for pandemic influenza” (Health Canada, 2000). Several attempts are undertaken to bring pandemic planning to multinational or even global level, such as the November 2001 conference on “Preparedness planning in the EU: Influenza and other health threats” and the WHO influenza pandemic preparedness plan (World Health Organization, 1999). Although some regions, e.g. the Canadian province of Ontario, are quite well advanced in pandemic planning and implementation, most of both national and international plans are not completed, let alone realized. Pandemic planning is a multifactorial process with a consequent high complexity. It is further complicated by the unpredictability of the next pandemic; its time of emergence, spread and severity cannot be foreseen and the efficacy of interventions available for annual influenza epidemics cannot be extrapolated per se to a pandemic situation. Both the complexity and the uncertainties surrounding pandemic preparedness have so far kept the various stakeholders from joining forces and tackling the problem from its roots. The complexity of preparedness planning must be simplified by dividing it into several “sub-projects”, which should be addressed separately to bring pandemic planning from abstract concepts to the desired concreteness. We developed a mathematical model, which shows the tangible consequences of conceptual plans by linking possible pandemic scenarios to health economic outcomes of possible intervention strategies. This model helps to structure the discussion on pandemic preparedness by calculating costs and benefits, gauges the public health benefits of optimized preparedness, and facilitates the translation of pandemic planning concepts to concrete plans.
2. Materials and methods A computer-based simulation model has been developed by MAPI Values (Macclesfield, United Kingdom) and Solvay Pharmaceuticals (Weesp, The Netherlands and Hannover, Germany), which allows for epidemiological, cost and efficacy inputs as well as manufacturing con-
straints. The model combines a vaccine production model with an adapted and revised cost-effectiveness model of influenza intervention strategies, based on that developed by Scuffham and West (2002). 2.1. Inputs for pandemic scenarios The model inputs for the pandemic scenarios determine the characteristics of the pandemic to which the several intervention strategies can be directed; these input parameters are the attack and mortality rate of that particular pandemic virus, the time of emergence of the pandemic virus and the time for the pandemic to spread and end. 2.2. Inputs for intervention scenarios The model inputs for intervention scenarios determine the availability of possible interventions, e.g. pandemic vaccine and antivirals, such as amantadine or the new neuraminidase inhibitors. These input parameters include specific vaccine production inputs, such as time to produce a suitable pandemic virus seed, time to produce a batch of vaccine, availability of eggs for egg-based vaccine production and available vaccine manufacturing capacity. They do not include specific antiviral production inputs, as it is assumed that if antivirals are selected as intervention strategy, they will be stockpiled and readily available. Intervention scenario inputs also include parameters, such as pandemic vaccine potency and dosing regime. It is assumed that all manufactured vaccines are used. 2.3. Inputs for intervention effectiveness Inputs for intervention effectiveness include parameters such as reduction of influenza-like illness (ILI) cases, reduction in primary care physician (PCP) visits for ILI symptoms, reduction of hospitalizations due to influenza and pneumonia, other respiratory illnesses and congestive heart failure. 2.4. Health economic inputs The model allows direct costs, from the perspective of the health care system, to be put in. Parameters for direct costs include all medical costs of the burden of the pandemic and for the intervention strategy selected. The medical costs for the burden of the pandemic include input parameters like costs of PCP consultations, costs of hospitalization due to influenza and pneumonia, other respiratory illnesses and to congestive heart failure. By combining average costs with resource use (PCP consultations, hospitalizations, etc.), the total medical costs for that particular pandemic can be calculated. The medical costs for the intervention strategy include the vaccine and vaccination cost as well as the costs of antiviral prophylaxis or therapy. The model allows for a series of secondary input parameters, such as the probability of having an antiviral prescribed
J.K. Medema et al. / Virus Research 103 (2004) 9–15
during a PCP visit with ILI symptoms, proportion of PCP visits undertaken at home and days off work per case. 2.5. Approach One way of giving structure to conceptual pandemic preparedness plans is to evaluate these plans from a health economic perspective. Therefore the developed model is used to calculate the costs and benefits of certain intervention scenarios for different pandemic scenarios. The strength of the model is its ability to visualize different pandemic scenarios and different preparedness strategies, which is reinforced by assuming the pandemic threat is imminent, rather than taking place at an unknown point in time decades from now. Hence we assumed the next pandemic strain emerges 1 January 2004, in order to show the current level of pandemic preparedness and how the model can visualize this. We assumed that the pandemic scenario caused by the “2004 pandemic” virus is based on the 1918–1919 pandemic with influenza attack rates of 20–50% depending on region and a case fatality rate of 1.5% in two pandemic waves (Potter, 1998). For the “2004 pandemic” an average 1918–1919 attack rate of 35% in a single wave with a case fatality rate of 1.87% (Scuffham and West, 2002) are put into the model as pandemic scenario parameters. Health economic parameters put into the model are based on average data obtained in the elderly in the United Kingdom in recent annual influenza epidemics (Scuffham and West, 2002). For the “2004 pandemic” case study it is assumed that health services utilization during the influenza pandemic are comparable to those during annual epidemics; increased absenteeism of health care workers and societal disruption by the pandemic are not taken into account. It is acknowledged that health care costs and health economical consequences of influenza in the United Kingdom cannot be translated to other countries, especially not to developing countries. Therefore, the latter are left out of this case study and the results that come from this study only visualize the level of preparedness of the developed countries. The total population of the developed countries is estimated at 1 billion. The key input parameters for the model, including vaccination efficacy, are listed in Table 1.
3. Results The outputs of the model that show the burden of the assumed “2004 pandemic” on developed countries are listed in Table 2 under the no intervention scenario. These data show that the no intervention scenario leads to 350 million cases and 22.2, 48.2 and 6.55 million influenza-related PCP consultations, hospitalizations and excess deaths, respectively. Total direct medical care costs add up to 166.6 billion, with hospitalizations costs of 165.2 billion as the major contributor.
11
Table 1 Key input parameters used within the model Vaccine production Month of pandemic strain emerging Month egg seed released by WHO Month last vaccine required in market Number of weeks to produce a vaccine Weeks cell seed released before egg seed Size of (trivalent) vaccine unit (g) Number of shots required per person Dose per shot (g) Annual manufacturing capacity
January 2004a,b April 2004a July 2004a,b 5a,b 9b 45a,b 2a,b 7.5a,b 230,000,000a,b
Demographics Overall population Target population Influenza attack rate (%) Years per life lost
1,000,000,000a,b 1,000,000,000a,b 35a,b 5c
Event probabilities PCP consultations (for influenza) Hospitalization rates due to: Influenza + pneumonia Other respiratory disease Congestive heart failure Mortality rate Vaccine effectiveness Vaccine efficacy (%) Reduction in GP visits (%) Reduction in hospitalizations due to: Influenza + pneumonia (%) Other respiratory disease (%) Congestive heart failure (%) Reduction in mortality (%) Unit costs and economic parameters PCP consultation ( ) Influenza vaccine price ( ) Influenza vaccine service and administration ( ) Discount rate Hospitalization for: Influenza + pneumonia ( ) Other respiratory disease ( ) Congestive heart failure ( ) Average hospital cost per bed day ( )
6.34d 2.52d 9.34d 1.9d 1.87d 53e 46f 39f 32f 27f 50f 28.6c 10a,b 3.1c 5c 3585c 3362c 3556c 261c
a
Authors’ assumption for egg-based vaccine intervention strategies. Authors’ assumption for cell culture-based vaccine intervention strategies. c Scuffham and West (2002). d Derived from Scuffham and West (2002). e Govaert et al. (1994). f Nichol et al. (1998). b
To visualize the current level of pandemic preparedness, the intervention strategies that can be launched to decrease the burden of the “2004 pandemic” are based on existing egg-based influenza vaccine technologies. Currently, the estimated Northern Hemisphere influenza vaccine usage is 230 million trivalent doses, which translates to a global influenza vaccine manufacturing capacity of 22 million monovalent doses (i.e. 15 g of pandemic strain antigen) per week. Two main starting materials are needed to manufacture the pandemic vaccine with the current technology: a sufficient amount of embryonated hen’s eggs bred under controlled
12
J.K. Medema et al. / Virus Research 103 (2004) 9–15
Table 2 Burden of different intervention scenarios in the assumed pandemic
Vaccination coverage (%) Number vaccinated Number of cases (influenza) PCP visits for treatment (influenza) Hospitalizations Excess deaths Discounted years per life lost PCP visit costs—vaccinations ( ) PCP visits costs—treatment ( ) Hospitalization costs ( ) Strategy costs ( ) Medical care costs ( ) Total direct costs ( )
No intervention
Intervention scenario 1 Egg-based vaccine manufacture
Intervention scenario 2 Cell culture-based vaccine manufacture
0 0 350,000,000 22,190,000 48,160,000 6,545,000 25,640,894 0 1,007,181,910 165,170,880,000 0 166,647,295,200 166,647,295,200
17 170,000,000 320,250,000 20,454,742 45,491,663 5,988,675 23,461,418 2,227,000,000 928,420,285 156,010,312,598 2,227,000,000 157,371,272,130 159,598,272,130
37 370,000,000 285,250,000 18,413,262 42,352,443 5,334,175 20,897,328 4,847,000,000 835,759,549 145,233,174,478 4,847,000,000 146,458,303,811 151,305,303,811
conditions and a seed of the virus strain to which the vaccine is directed. Preparation of the latter is performed by WHO collaborating centers and takes 1.5–2.5 months for annual epidemic strains. For the assumed “2004 pandemic” seed preparation time is unknown and unpredictable; we assumed a seed preparation time of 3 months, which means that pandemic vaccine manufacture can start 1 April 2004. This is within the regular annual influenza vaccine manufacturing season, which means that the eggs are readily available for pandemic vaccine manufacture. We assumed that the applied “2004 pandemic” vaccine will be 7.5 g monovalent given in two doses and should be administered within 6 months after emergence of the “2004 pandemic” virus. The results of the assumed intervention strategy on the “2004 pandemic” are listed in Table 2 as intervention scenario 1. These data show that the current level of pandemic preparedness in the assumed egg-derived scenario leads to vaccination of 17% of the population, which avoids 29.8 million influenza cases, 1.74 million PCP visits, 2.67 million hospitalizations and 556,000 deaths compared to a scenario of no vaccination. The intervention strategy costs add up to 2.2 billion, but as this strategy leads to a reduction of 9.3 billion medical care costs, the saving on direct costs of the “2004 pandemic” is 7.1 billion. The question for pandemic planners is if the level of pandemic preparedness can be improved and which intervention strategies should be selected to achieve this. As vaccination is the most cost-effective intervention, an improved intervention strategy from a health economic perspective would be to produce more pandemic vaccine in the available time frame. This can be achieved by increasing the time period for vaccine manufacture, by increasing the vaccine manufacturing capacity or preferably by a combination of the two. An earlier availability of a suitable seed virus would increase the time frame for pandemic vaccine manufacture and hence improve the intervention strategy. The new cell culture-based influenza vaccine manufacturing technology likely facilitates seed preparation, as mammalian cell culture may require less adaptation of viruses and allows con-
tainment of more pathogenic pandemic viruses without prior decrease of their virulence (see Section 4). To visualize this we calculated the level of pandemic preparedness with an intervention scenario of cell culture-based vaccine manufacturing, assuming that preparation of a cell culture-based seed virus takes 1 month and assuming that the current global weekly manufacturing capacity of 22 million monovalent doses is cell culture-based in stead of egg-based. The results of this assumed intervention strategy on the “2004 pandemic” are listed in Table 2 as intervention scenario 2. These results show that, compared to no intervention scenario, the cell culture-based intervention strategy avoids 75 million influenza cases, 3.78 million PCP consultations for influenza treatment and, respectively, 5.81 million and 1.21 million influenza-related hospitalizations and excess deaths. Compared to the assumed egg-based vaccine intervention with 17% vaccine coverage, the cell culture-based intervention strategy leads to vaccination of 37% of the population, avoiding an additional 35 million influenza cases, 2.04 million PCP consultations for influenza treatment, 3.14 million influenza-related hospitalizations and 654,500 excess deaths. The cell culture-based intervention strategy costs add up to 4.9 billion, but as this strategy leads to a reduction of 20.2 billion medical care costs, the saving on direct costs of the “2004 pandemic” is 15.3 billion, an additional 8.3 billion compared to the egg-based intervention scenario. The years per life lost gained are 2.56 million, whereas the cost per life year gained is 3198. The cost per case, hospitalization and death averted are, respectively, 234, 2612 and 12,530.
4. Discussion The “2004 pandemic” case study is based on numerous assumptions. Each can be challenged and, indeed, such uncertain variables are discussed by pandemic planners to challenge preparedness concepts. However, the assumed parameters are only inputs to the presented model and not part of
J.K. Medema et al. / Virus Research 103 (2004) 9–15
the model itself. Therefore, questioning these assumptions is not questioning the model itself. The model is only a tool to visualize the consequences and effectiveness of a number of planning concepts, to which it lends structure and logic. Moreover, a sensitivity analysis of the model can be used to set priorities within pandemic planning discussions; the more sensitive the model is to a given assumption, the higher the effect of that particular input on the level of pandemic preparedness. A sensitivity analysis of the “2004 pandemic” case study shows that the model is most sensitive to the pandemic scenario in terms of attack and mortality rates, as well as to the intervention scenarios in terms of effectiveness and available amounts of pandemic vaccine. It is less sensitive to medical care costs and relatively insensitive to pandemic vaccine price. The total direct medical costs of 166.6 billion are remarkably close to the 166.5 billion calculated by the CDC for a pandemic with 35% attack rate (Meltzer et al., 1999). However, these are for the United States alone with 89 million cases, 734,000 hospitalizations and 207,000 excess deaths on a smaller population than in the model presented here. Although the CDC model is similar, the inputs are based on US data, explaining the higher costs. 4.1. Pandemic scenario The assumptions put into the model to derive the presented “2004 pandemic” scenario have been based on the 1918–1919 pandemic in terms of spread and severity. The first cases were reported in March 1918, leading to a first wave that was not severe. It took about 5–6 months before the pandemic evolved into a second wave with the particular characteristics of high attack and mortality rates, in the age groups at risk for epidemic influenza as well as other age groups, like children and young adults (Potter, 1998). This pandemic shows that everybody can be at risk for pandemic influenza and means that at that time intervention in all age groups at risk should have taken place in these 6 months between emergence and actuality of the pandemic. However, both knowledge and surveillance of circulating influenza viruses have been well established over the last 50 years, enabling an earlier identification of potential pandemic viruses and consequently an earlier onset of intervention programs compared to 1918. Furthermore, improved influenza diagnosis and the availability of intervention enable us to significantly reduce pandemic virus spread. On the other hand, the increased globalization and international travel in current times are likely to speed up spread of pandemic viruses to a large extent. As these phenomena counteract, we assumed for the “2004 pandemic” the same time frame between emergence and spread of the pandemic virus as in 1918. The severity of a pandemic is reflected in the model by attack rate and event probabilities, such as influenza-related PCP consultations, hospitalizations and case fatality rate. For the “2004 pandemic” we assumed a higher attack rate
13
to reflect a higher severity compared to annual epidemics. All event probabilities for the “2004 pandemic” are based on annual epidemic UK event probabilities for the elderly, which are higher than for other age groups. However, the higher attack rate implies that during the “2004 pandemic” other age groups are affected as well and extrapolation of high event probabilities in the elderly to these age groups also reflects a more severe event. 4.2. Egg-based vaccine intervention scenario Health economic studies on annual influenza epidemics have shown that influenza vaccination is by far more cost-effective than prescription of antivirals for prophylaxis or therapy (Scuffham and West, 2002). Because the health economic perspective has been used in the model to visualize the current level of pandemic preparedness, vaccination has been selected as the pandemic intervention strategy in the current case study as well. The case study is limited to the developed world, because the health economic inputs are based on data gathered in a developed country, but pandemics are a global event and the burden in developing countries might be even greater because of lower public health. However, as long as global pandemic preparedness plans have not been realized, the world will face a limited availability of interventions. These will be made available to developed countries first at the cost of the developing world. The intervention scenario input parameters are based on current influenza vaccine manufacturing technologies. We assumed that the applied “2004 pandemic” vaccine will be 7.5 g monovalent given in two doses, as the human population is immunologically na¨ıve to the pandemic strain and will need a booster-dose to elicit a protective immune response. Clinical studies in unprimed individuals indeed indicate that a two dose regime is needed (Nicholson et al., 1979), but that this may be achieved with a lower, possibly adjuvanted dose (Nicholson et al., 2001; Hehme et al., 2002). The next pandemic is likely to be detected by the WHO global influenza surveillance network and, just as for annual epidemic strains, WHO collaboration centers prepare the pandemic virus seed. Seed preparation time for the next pandemic is unpredictable and cannot be based on experience with epidemic strains; although the H5N1-subtype that emerged in Hong Kong in 1997 appeared only a candidate pandemic virus, a suitable seed is still not available 5 years later. In case of a high pandemic threat the efforts that will be put into seed preparation will be accordingly and preparation times for pandemic seeds suitable for egg-based vaccine manufacture in general is expected to be 2–8 months (internal communication J. Woods of WHO Collaboration Center NIBSC, WHO Experts-IVS Taskforce Meeting, 26 September 2002, Geneva). For the case study we assumed 3 months, leaving 3 months for vaccine manufacture, distribution and vaccination before the “2004 pandemic” would evolve and widely spread.
14
J.K. Medema et al. / Virus Research 103 (2004) 9–15
4.3. Cell culture-based vaccine intervention scenario Field strains circulating in the human population usually do not propagate well on eggs and need to be attenuated for egg-based vaccine manufacture. Therefore, both epidemic and pandemic seed preparation for egg-based technology mainly consists of preparation of so-called high growth reassortants by reassortment of field strains with strains that propagate well on eggs. Additionally, pandemic viruses are more virulent than epidemic strains and need to be attenuated to decrease their virulence to safely propagate such viruses in existing egg-based manufacturing facilities, as these are based on open systems. On the contrary, cell culture-based vaccine manufacture is performed in closed bioreactor systems, enabling the desired containment to process more virulent pandemic viruses without prior attenuation and hence less laborious seed preparation. Moreover, cell culture-based influenza vaccine manufacturing technologies make use of mammalian cell lines, such as MDCK, which are more closely related to the human host than the avian egg-based technology. MDCK-grown human influenza field isolates indeed are generally more antigenically homogenous and more alike the original field strain than egg-grown isolates (Zambon, 1998), which is probably caused by host cell selection of variants. Mammalian epithelioid cells like MDCK appear to be the most sensitive cell culture system for human influenza viruses to date (Zambon, 1998), whereas some human influenza isolates need to be adapted to growth in the allantoic cavity (Murphy and Webster, 1996). Therefore, cell culture is generally used for primary isolation of human influenza viruses. The probability that human pandemic viruses readily grow in mammalian cells therefore seems higher than in the more distant avian host, thereby making the preparation of high growth reassortants obsolete and decreasing the time for pandemic seed preparation. This favors the preparation of pandemic seeds by propagation of the pandemic strain directly on mammalian cell culture in stead of via embryonated eggs. However, for annual epidemic vaccine manufacture the egg-passage is a regulatory obligation; omission of this procedure therefore needs to be cleared by authorities. As the time period for vaccine manufacture is mainly determined by the time a suitable virus seed is available and the time the pandemic spreads in the population, a potential quicker release of pandemic seed suitable for cell culture enables the production of more vaccines. The more vaccines, the greater the vaccination rate and the greater the opportunity to benefit from both reduced mortality and reduced total costs. This underlines the importance of cell culture-based influenza vaccine manufacturing as tool for increased pandemic preparedness. By assuming the pandemic virus emerges in January, eggs are readily available for pandemic vaccine manufacture once the seed is available. However, a pandemic virus can emerge all year round, also outside the annual epidemic vaccine manufacturing period. Given that logistics of egg
preparation require ordering about 1 year in advance, availability of sufficient eggs outside the planned period certainly can be questioned. Cell culture-based vaccine manufacture however makes long-term advance planning obsolete, as all starting materials are in stock and readily available whenever required. This advantage however is not illustrated in the assumed “2004 pandemic”, thereby underestimating the advantageous health economics of cell culture-based over egg-based intervention and thus the importance of cell culture-based vaccine manufacturing as tool for increased pandemic preparedness. Although the presented model can lend structure and logic to pandemic preparedness discussions, pandemic planning remains a complex process. The efficacy of interventions, such as vaccines and antivirals, during pandemics needs to be studied and proven beforehand as much as possible in order to make such interventions available in time. Regulatory procedures of such interventions need to be harmonized and adapted to the short-time lines available in pandemic situations and logistical systems need to be set up or streamlined to successfully execute the intervention strategy. Pandemic vaccination might be the most cost-effective approach, but will require the availability of a suitable virus seed and adequate manufacturing facilities to process such a seed. In order to manufacture sufficient amounts for an adequate level of pandemic vaccine, manufacturing capacity needs to be increased. As stated by WHO, policy makers need to keep in mind the several years needed to construct new production facilities and significantly increase production capacity (World Health Organization, 2002). This only has an economical incentive if interpandemic vaccine usage is increased, which—just as pandemic preparedness—is a joint responsibility of the public and private sector.
Acknowledgements The authors like to thank James Piercy of Mapi Values for the health economic support during development of the model and Professor Kristin L. Nichol of the University of Minnesota Medical School for her valuable comments on the manuscript.
References Davies, P., 2000. Catching Cold, 1918s Forgotten Tragedy and the Scientific Hunt for the Virus that Caused It. Penguin Books, Harmondsworth, UK, 306 pp. Govaert, T.M., Sprenger, M.J., Dinant, G.J., Aretz, K., Masurel, N., Knottnerus, J.A., 1994. Immune response to influenza vaccination of elderly people. A randomized double-blind placebo-controlled trial. Vaccine 12, 1185–1189. Health Canada, 2000. Canadian Contingency Plan for Pandemic Influenza—Draft January 5, 2000. The Laboratory Centre for Disease Control, Health Canada, Ottawa.
J.K. Medema et al. / Virus Research 103 (2004) 9–15 Hehme, N., Engelmann, H., Kunzel, W., Neumeier, E., Sanger, R., 2002. Pandemic preparedness: lessons learnt from H2N2 and H9N2 candidate vaccines. Med. Microbiol. Immunol. 191 (3–4), 203–208. Meltzer, M.I., Cox, N.J., Fukuda, K., 1999. The economic impact of pandemic influenza in the United States: priorities for intervention. Emerg. Infect. Dis. 5, 659–671. Murphy, B.R., Webster, R.G., 1996. Orthomyxoviruses. In: Fields, B.N., Knipe, P.M., Howley, P.M. (Eds.), Fields Virology, 3rd ed. Lippincott-Raven, Philadelphia, pp. 1397–1445. Nichol, K.L., Wuorenma, J., Von Sternberg, T., 1998. Benefits of influenza vaccination for low-, intermediate-, and high-risk senior citizens. Arch. Intern. Med. 80, 1769–1776. Nicholson, K.G., Tyrrell, D.A.J., Harrison, P., 1979. Clinical studies of monovalent inactivated whole virus and subunit A/USSR/77 (H1N1) vaccine: serological responses and clinical reactions. J. Biol. Stand. 7, 123–136. Nicholson, K.G., Colegate, A.E., Podda, A., Stephenson, I., Wood, J., Ypma, E., Zambon, M.C., 2001. Safety and antigenicity of non-adjuvanted and MF59-adjuvanted influenza A/Duck/Singapore/97 (H5N3) vaccine: a randomized trial of two potential vaccines against H5N1 influenza. Lancet 357, 1937–1943. Patriarca, P.A., Strikas, R.A., Gensheimer, K.F., Cox, N.J., Fukuda, K., Meltzer, M.I., 1999. A pandemic influenza planning guide for
15
state and local officials. Draft 2.1—National Vaccine Program Office, Centers for Disease Control and Prevention, Atlanta (available at http://www.cdc.gov/od/nvpo/pub.htm). Potter, C.W., 1998. Chronicle of influenza pandemics. In: Nicholson, K.G., Webster, R.G., Hay, A.J. (Eds.), Textbook of Influenza. Blackwell Science, Oxford, pp. 3–18. Scuffham, P.A., West, P.A., 2002. Economic evaluation of strategies for the control and management of influenza in Europe. Vaccine 20, 2562–2578. Strikas, R.A., Wallace, G.S., Myers, M.G., 2002. Influenza pandemic preparedness action plan for the United States: 2002 update. CID 35, 590–596. World Health Organization, 1999. Influenza Pandemic Preparedness Plan: The Role of WHO and Guidelines for National and Regional Planning. WHO/CDS/CSR/EDC/99.1, Geneva (available at http://www. who.int/emc-documents/influenza/whocdscsredc991c.html). World Health Organization, 2002. Draft WHO Guidelines on the Use of Vaccines and Antivirals during Influenza Pandemics. Issued 2–4 October 2002, Geneva (available at http://www.who. int/emc/diseases/flu/whoguidelines.htm). Zambon, M., 1998. Laboratory diagnosis of influenza. In: Nicholson, K.G., Webster, R.G., Hay, A.J. (Eds.), Textbook of Influenza. Blackwell Science, Oxford, pp. 3–18.