The role of economic evaluation in vaccine decision making

Abstract

In recent years, several countries have experienced increases in the incidence of serogroup C meningococcal disease. It can be controlled with older polysaccharide vaccines and particularly the recently developed conjugate vaccines. For 21 developed countries, we investigated the role that economic evaluation played in the decision to introduce the conjugate vaccine into either the routine childhood vaccination schedule, as a mass vaccination ‘catch-up’ campaign or not at all. A literature review was performed and experts from these countries were contacted.

For six countries, we identified published economic evaluations for meningococcal C conjugate vaccination. In four of them (Australia, Canada [Quebec], The Netherlands and the UK) the analyses were performed before a decision about the use of the conjugate vaccine was made. In all of these countries, the economic evaluation offered guidance as to the most efficient way to add the conjugate vaccine to the routine infant immunisation schedule and, in three countries, this advice was adopted by decision makers. In Portugal and Switzerland, initial vaccination decisions were made without the economic evaluations that are influencing current decision making. Of the countries without economic evaluations, six implemented vaccination programmes. Overall, there was a positive correlation between the reported incidence of meningococcal C disease and (a) the decision to vaccinate and (b) performing an economic evaluation.

All economic evaluations were modelling studies. The reported cost-effectiveness ratios were sensitive to the age of vaccination, the future meningococcal incidence, vaccine price and some methodological characteristics that varied widely between studies making direct comparisons difficult.

In conclusion, in almost all countries where economic evaluations for meningococcal C conjugate vaccinations have been conducted, their results had an important role in the decision-making process. However, in most countries with strongly increasing meningococcal incidence, public health considerations took precedence. In order to improve the international comparability of such studies, firmer national and international modelling guidelines and better adherence to such guidelines seem necessary.

Neisseria meningitidis causes significant mortality and morbidity worldwide. In recent years, the proportion of meningococcal disease cases due to serogroup C has increased in several industrialised countries, in Europe (Austria, Belgium, Czech Republic, France, Iceland, Ireland, Portugal, Spain, The Netherlands, the UK)[1] as well as in Australia[2] and Canada,[3] triggering some countries to launch large-scale meningococcal C vaccination campaigns and/or routine vaccination. As a result of their size, these programmes require very high investments, so an assessment of their cost effectiveness seems highly desirable before they are started.

In this article, we will investigate the role of economic evaluations in vaccine decision making with a focus on the new meningococcal serogroup C conjugate (MCC) vaccines in developed countries. First, we briefly describe the pathogen and possible prevention measures. Second, we present the results of published economic evaluations for meningococcal C vaccination and identify the main factors that determine the cost effectiveness of such programmes. Third, we show the role that economic modelling studies have played in the decision for or against the introduction of MCC vaccination strategies in several countries.

1. Epidemiology

N. meningitidis is an exclusively human, gram-negative diplococcus that is spread from person to person by nasopharyngeal secretions of carriers or patients. Transmission typically requires close contact as N. meningitidis usually dies quickly outside of the human body. During periods of endemic infection, 5–10% of the population carry the bacteria asymptomatically in their nasopharynx.[4] Carriage is lowest in young children and highest in adolescents and young adults. However, only a small proportion of those colonised develop invasive meningococcal disease, where the bacteria spreads through the bloodstream and to the brain.

Meningococci are classified into serogroups based on antigenic differences in the capsular polysaccharide. There are 13 meningococcal serogroups, but the most invasive disease is caused by serogroups A, B, C, Y and W-135. Serogroup A is responsible for most cases of meningococcal disease in the sub-Saharan African zone known as the ‘meningitis belt’, while serogroups B and C cause most disease in developed countries. In Latin America, serogroup B is most common while in the US, serogroup Y is an important cause of disease, in addition to serogroups B and C. Meningococcal disease is seasonal; most cases typically occur in the ‘meningitis belt’ at the end of the dry season, in Brazil between the rainy and dry season, and in the US and Western Europe between late winter and early spring.[4–9]

The incidence of serogroup C meningococcal disease (SCMD) is highly age specific, peaking between age 0–2 years with a secondary peak in disease incidence in teenagers aged between 14 and 18 years. Table I shows the overall reported incidence in selected countries.

Table I

Laboratory confirmed cases of serogroup C meningococcal disease (SCMD) in selected countries^a

In 2002, the overall case fatality rate from reported SCMD cases in Europe was 12%, although this varied by age, with much higher rates in adults.[1] Serious sequelae such as amputations and scars (due to skin and limb necrosis), hearing loss, renal failure (due to muscle necrosis) and neurological sequelae (developmental delay, local neurological deficits and seizures) can occur in 3–15% of survivors.[4,28–31] Some studies have shown that SCMD, particularly those cases caused by the hypervirulent C2a strains that have predominated recently, are associated with increased mortality[32] and morbidity[33] compared with serogroup B disease.

The high case fatality and sequelae rates, even in countries with advanced healthcare systems, are two important features of SCMD that increase public perception and raise the public health importance of this disease. Disease progression can be rapid, and early symptoms may be difficult to recognise. Suspected cases should be treated immediately with antibacterials.[4,8,9] Although most cases of SCMD are sporadic, outbreaks of SCMD requiring intensive public health management occur occasionally.

2. Prevention of Serogroup C Meningococcal Disease

2.1 Chemoprophylaxis

Chemoprophylaxis (e.g. rifampicin, ceftriaxone or ciprofloxacin) is recommended for all close contacts of persons with SCMD and should be taken as soon as possible after the most recent contact with the index case.[34–39] Household members and room-mates have a 1000-fold increased risk and pre-primary school contacts a 50-fold increased risk of acquiring SCMD.[4]

2.2 Vaccination

2.2.1 Polysaccharide Vaccines

The first polysaccharide vaccines against serogroups A and C were developed over 30 years ago. Today, polysaccharide vaccines containing the purified polysaccharide capsules of group A and C (bivalent) or A, C, Y and W-135 (quadrivalent) are licensed and available worldwide.[40] These vaccines are not immunogenic in children aged <2 years and vaccine efficacy is highly age dependent, increasing with age through childhood. As polysaccharides are thymus-independent antigens, they induce no immunological memory and thus protection from vaccination is limited to approximately 3–5 years. In addition, after repeated vaccination, immune hyporesponsiveness to meningococcal C polysaccharide has been observed.[5] As a result of these shortcomings, the suggested use of these vaccines has been limited to the control of outbreaks of SCMD and for prophylaxis of persons at increased risk, e.g. contacts of SCMD index cases, travellers to areas with epidemic or hyperendemic SCMD, persons with immune deficiencies and laboratory workers with exposure to N. meningitidis.[35,37,39]

2.2.2 Conjugate Vaccines

In 1999, the first MCC vaccine was licensed in the UK. Since then, several other countries have also licensed MCC vaccines. In these vaccines, purified polysaccharides of capsule C are covalently conjugated to a protein carrier (Cross Reacting Material 197 or tetanus toxoid).[41] As a result, they cause a thymus-dependent immune response and thus induce immunological memory. These vaccines have been found to be immunogenic in persons of all ages, including young infants. UK data suggested a short-term effectiveness of >90%.[42–45] However, a recent study suggests that the duration of protection is highly dependent on the age at vaccination, being much shorter for infants immunised according to the accelerated 2-, 3- and 4-month schedule.[46] Another study showed rapidly waning seroprotection in children that received a single MCC vaccine dose at a median age of 2.3 years.[47] As expected following the experience with Haemophilus influenzae type b (Hib) vaccines,[48] MCC vaccines have been shown to reduce the prevalence of carriage in targeted age groups,[49] and significant levels of indirect protection (herd immunity) have been observed.[50] Immune hyporesponsiveness has not been reported for MCC vaccines.[51]

An assessment of the long-term effects of these vaccines is difficult as the MCC vaccines have just been introduced. For instance, we are unclear about the level of waning immunity and whether booster doses are necessary. There have been concerns that selective vaccination against just one serogroup might lead to capsule switching among meningococcal strains and strain replacement by other meningococci.[9] Reports from Canada, the Czech Republic and Spain suggest that capsular switching has occurred to some extent,[52] but careful assessment of disease strains in the UK has not revealed any evidence of this to date.

Currently in the UK, children aged up to 6 months receive three vaccine doses at age 2, 3 and 4 months, while persons aged ≥1 year receive one dose only. New data suggest that the number of MCC vaccinations used in the infant schedule might be reduced to two or even one as vaccines with improved immunogenicity become available.[53]

Bivalent (against the serogroups A and C) and quadrivalent (A, C, Y and W-135) conjugate vaccines are currently being developed, as well as combination vaccines (e.g. meningococcal C and pneumococcal vaccines).[40] A quadrivalent conjugate vaccine was approved by the US FDA in January 2005.[54] Furthermore, as part of the Meningitis Vaccine Project,[55] a monovalent serogroup A conjugate vaccine is also being developed.

3. Determinants of Cost Effectiveness

We screened the databases PUBMED and EMBASE for relevant articles using the search terms ‘meningococcal or meningococci or meningitidis’ combined with the search terms ‘economic evaluation or costs or cost’. Furthermore, we searched the Health Economic Evaluation Database (HEED) and the UK NHS Economic Evaluation Database (NEED) by using the terms ‘meningococcal’ or ‘meningococci’ or ‘meningitidis’. All database searches were not limited to a date range and were performed in October 2003. Abstracts from the retrieved studies were screened. In addition, abstracts from conferences found through hand searching were also included. We also contacted experts from the European Union Invasive Bacterial Infections Surveillance (EU-IBIS) Network (http://www.euibis.org) and the European Meningococcal Network Study (EUMenNet) [http://neisseria.org] and asked: (a) whether an MCC vaccination programme had been implemented in their country; and (b) whether they were aware of any economic evaluation that supported the decision process in their country. Using this approach, we gathered information on 21 countries as listed in table I.

We identified 13 published economic evaluations[22,54,56–67] for meningococcal C vaccination programmes, of which three investigated the cost effectiveness of vaccination programmes in the US with the quadrivalent polysaccharide or quadrivalent conjugate vaccine. We excluded these three studies as the serogroups W-135, and especially Y, substantially contribute to the burden of meningococcal disease in the US, and the results were only given for the total number of prevented meningococcal cases.[54,59,63] On the other hand, an Australian study[64,65] that also evaluated the tetravalent vaccine was included as the serogroups A, Y and W-135 play no important role in Australia. We also identified two unpublished studies for the UK: one by two of the current authors[68] and one by the UK Department of Health. The latter was unavailable and so was excluded.

Table II and table III present the main results and the applied methodology of the included economic evaluations. The average cost-effectiveness ratio (CER) shows the costs per gained health effect of a vaccination strategy compared with a ‘no vaccination strategy’, while the incremental cost-effectiveness ratio (ICER) presents the costs of each additional health effect compared with a less expensive but also less effective vaccination strategy. Clearly, if there are different, mutually exclusive vaccination strategies, the incremental and not the average CER should be used. For example, the average CER for routine MCC vaccination at 2, 4 and 6 months in Switzerland has been estimated at approximately €55 000 (year 2002 values) per QALY gained.[62] However, the estimated ICER of this programme versus one-dose routine vaccination at 12 months is €589 000 per QALY gained, i.e. each additional QALY that is gained by vaccinating at 2, 3 and 6 months versus at 14 months costs €589 000 (table II).

Table II

Economic evaluations of routine meningococcal C vaccination^a

Table III

Economic evaluations of once-only mass vaccination campaigns^a

All of the economic evaluations we identified were modelling studies. The importance of different models and model parameters is shown in table IV and summarised below.

Table IV

Reported sensitivity of the cost-effectiveness ratio of meningococcal vaccination to variations in different model parameters reported in the reviewed economic evaluations

Because of the age-specific incidence of SCMD, the age of vaccination has a strong impact on the outcomes of vaccination. Furthermore, vaccinating children before the age of 1 year renders a higher CER than after that age because it requires three (vaccinating at age 2, 3 and 4 months, as in the UK vaccine schedule) or two (at age 5 and 6 months) vaccine doses compared with one dose for persons aged ≥1 year.

A major determinant of cost effectiveness for any of the meningococcal C vaccination strategies is the assumed degree and duration of protection in vaccine recipients. Short-term estimates suggest that MCC vaccine effectiveness is high, but because of limited experience with this vaccine, the assumptions regarding the duration of protection are still very uncertain. Recent study results suggest that the duration of protection is highly dependent on the age at vaccination, waning much faster after vaccination at age 2, 3 and 4 months than after vaccination at older ages.[46] This information was used in the latest economic evaluation by Trotter and Edmunds[68] resulting in a higher effectiveness of routine childhood vaccination at the age of 12 months versus at 2, 3 and 4 months (table II). In general, a higher effectiveness and a longer assumed duration of protection (e.g. 20 instead of 5 years) leads to a more favourable CER.

The future incidence of SCMD is one of the most influential model parameters. Unfortunately, this estimate is uncertain because of unpredictable temporal trends in incidence. For this reason, most studies investigated the impact of changing incidence assumptions or epidemiological scenarios in the sensitivity analysis. Parameters related to disease progression, such as case fatality and the sequelae rates, are less uncertain. A higher future incidence of SCMD as well as higher case fatality and sequelae rates will render a more attractive CER.

The price of the vaccine and, to a lesser extent, the treatment costs of meningococcal sequelae can have an impact on the results. While a higher vaccine price leads to higher CERs, higher treatment costs will produce lower CERs.

The choice of perspective (governmental, healthcare payer, societal) can affect the CER. As vaccination requires immediate investment of costs but renders prevented sequelae and associated savings later, a higher discount rate for future costs and effects will result in a less attractive CER.

The inclusion of herd immunity improves the cost effectiveness of routine childhood immunisation. However, if a catch-up programme has been conducted, the resulting herd immunity effect will increase the ICER of routine childhood immunisation of children aged <1 years compared with immunisation at 1 year. The additional health gain and the associated savings of immunising children before the age of 1 year decreases, while the intervention costs stay the same. For The Netherlands, it has been estimated that if 50% of children aged <14 months are protected by herd immunity due to the catch-up programme, the incremental costs per life-year gained (LYG) for vaccination at 2, 3 and 4 months versus 14 months would almost double.[67] As all the published studies used static decision-analysis modelling, the herd immunity effects could only be estimated based on epidemiological data and not by simulating the disease spread. A dynamic model has been developed and used to reassess the cost effectiveness of meningococcal C vaccination in the UK, allowing the relative contributions of routine and catch-up vaccination to the herd immunity effects to be assessed (see table II).[68]

As SCMD frequently gives rise to long-term sequelae, the use of QALYs or disability-adjusted life years (DALYs) gained instead of LYG, i.e. the consideration of quality-of-life (QOL) changes, considerably improves the cost effectiveness of vaccination. For most countries, there are no specific QALY or DALY weights available for all complications and sequelae of SCMD and thus they were often ignored. In most studies that did include QOL weights, the estimates used should be seen as rough estimates only; there was a lack of high-quality data.[62,64,65,67,68] Furthermore, recent results indicate that the QALY weights of permanent sequelae after meningitis strongly depend on the applied classification method, i.e. whether the EuroQoL Five-Dimension Questionnaire or the Health Utility Index is used.[70] In general, methodology of QOL assessments for young children is still in its infancy. Clearly, this is an important area requiring further research.

Because of the high case fatality and long-term sequelae rates of SCMD, the chosen indirect cost approach can strongly influence the CER. The CER of routine MCC vaccination of Dutch children at 14 months decreased by 688% when the human capital approach instead of friction cost method was used.[67]

Comparison of the sensitive methodological characteristics in table IV with the applied methods in tables II and III demonstrates that there are strong differences in important model parameters between the studies, e.g. the discount rate and the measurement method for indirect costs.

4. Influence of Economic Evaluation on Decision Making

Table V shows the status of MCC vaccination in 21 countries in December 2003.

Table V

Meningococcal serogroup C conjugate (MCC) vaccination programmes and related economic evaluations in selected countries (December 2003)

In Austria, Denmark, France, Germany, Israel, Italy, Malta, Sweden and the US, neither an economic evaluation for MCC vaccination has been conducted nor an MCC campaign started. All of these countries have a rather low incidence of SCMD (table I) and are monitoring its development. In the US, an MCC vaccine has been approved for use in persons aged 11–55 years only.[54]

At the end of the 1990s, Iceland and Ireland had the highest incidence of SCMD. Both countries implemented a catch-up MCC vaccination campaign without performing an economic evaluation. In addition, Ireland introduced the MCC vaccine in its routine childhood vaccination schedule. In response to a significant increase of SCMD cases in early 2001, the Belgian government initially planned to commission an economic evaluation. However, before any research proposal was accepted, the decision to proceed with vaccination was taken in the spring of 2001 and the phased mass campaign started in November 2001. Thus, the decision to vaccinate was purely based on the increase of SCMD and likely sped up by the media attention it received. The decision in Greece and Luxembourg was made on similar grounds of increasing incidence of serogroup C compared with serogroup B disease.

4.1 Australia

In Australia, an unpublished model-based cost-effectiveness (costs per case, death, DALY averted and LYG) and cost-benefit analysis was prepared for the Australian Technical Advisory Group on Immunisation (ATAGI) in 2002. Based on this report, ATAGI recommended universal vaccination at age 12–13 months, with additional universal vaccination at age 15 years, as well as a one-off vaccination campaign for 16- to 17-year-olds. The incremental cost effectiveness of implementing a routine three-dose schedule starting at 2 months was not thought justifiable on the basis of this analysis. Mass vaccination of all cohorts between the ages of 1 and 18 years was preferred on the basis of effectiveness alone, but estimated to be the most expensive and the least cost-effective option. Nonetheless, considering that this last programme would prevent the most cases and deaths, the government chose to provide it, in phases as of early 2003, and initiate annual universal vaccination at age 12–13 months.

4.2 Canada

SCMD increased in five Canadian provinces (Alberta, British Columbia, Manitoba, Quebec and Ontario) between 1999 and 2001, triggering province-wide or local immunisation campaigns. High-school-aged children were the main target group, but also younger children and young adults were included to a varying extent.[18] Alberta (between 2000 and 2001)[78] and Quebec (in 2001)[79] responded to the SCMD increase by launching mass vaccination campaigns. As the MCC vaccine was only licensed in Canada in April 2001, Alberta mainly used the polysaccharide vaccine while Quebec used the MCC vaccine only. A study by the Comité sur l’immunisation du Québec (CIQ)[56] showed that using the MCC vaccine in children only and the polysaccharide vaccine in teenagers instead of using the MCC vaccine for the whole target population would be less effective but more cost effective for Quebec (table III). As the decision makers found it unacceptable to deny the population the best vaccine, the MCC vaccine was used for the whole target population.[73]

In 2001, the Canadian National Advisory Committee on Immunization recommended routine infant immunisation with three doses of MCC vaccine at age 2, 4 and 6 months and catch-up immunisation with two doses for infants aged 4–11 months and one dose for individuals aged ≥1 year.[41] In Alberta, routine MCC vaccination at age 2, 4 and 6 months was introduced in November 2001 and no formal cost-effectiveness analysis was performed. In Quebec, a cost-effectiveness analysis was requested by the Ministry of Health, with Health Canada also providing financial support.[58,80] Results indicated that the most cost-effective control strategy was a one-dose routine programme in most likely epidemiological scenarios (table II).[58] On this basis, a routine programme was implemented in November 2002, with one dose of conjugate vaccine being offered at age 12 months, along with the first measles, mumps, rubella (MMR) dose.[73] This economic study also influenced the decision to implement a similar programme in British Columbia, starting in July 2002, although here the vaccine was also offered to all unvaccinated schoolchildren in grade 6 (De Wals P, personal communication).

4.3 The Netherlands

Between 1993 and 2000, the reported annual number of cases of SCMD varied between about 60 and 110 in The Netherlands. In 2001, the number of reported SCMD cases increased to almost 280,[67] causing high anxiety in the population. The government therefore requested an economic evaluation of MCC vaccination.

The analysis showed that all investigated vaccination options would render substantial health gains. Both the catch-up programme and the routine childhood vaccination at 14 months were found to be cost effective from the societal and healthcare payer perspective, based on the €20 000 per LYG or QALY cut-off point for healthcare programmes in The Netherlands.[81] Other childhood vaccination options that were considered, most notably vaccination at age 2, 3 and 4 months or at 5 and 6 months, showed highly unfavourable ICERs when compared with vaccination at 14 months.[67]

Based on these results, in early 2002, the Dutch government made the decision to implement the catch-up vaccination programme and routine childhood vaccination at 14 months.[67]

4.4 Portugal

Since February 2003, the Portuguese government has reimbursed 40% of the costs of MCC vaccination. In January 2004, the National Pharmacy and Medicines Institute (Infarmed), a governmental agency, decided that the reimbursement of these costs should be continued (Silverio N, personal communication). This decision was influenced by the Portuguese economic evaluation[22] (Infarmed, personal communication). The National Vaccination Plan Committee has recently decided to introduce the MCC vaccine into the national routine childhood vaccination programme in 2005 and to perform a catch-up campaign.[76] The Portuguese economic evaluation (see table II)[22] may have had an influence on this decision as well.

4.5 Spain

In response to the increasing incidence of SCMD in Spain during the 1990s and the associated public concern, 14 of the 17 Spanish regions conducted mass vaccination campaigns with the meningococcal polysaccharide vaccine in 1997. In 1999, the vaccine effectiveness started to decrease and the incidence of SCMD began to increase. The health authorities of all Spanish regions then decided to include the MCC vaccine in the routine childhood vaccination schedule at age 2, 4 and 6 months.[82,83] In addition, catch-up immunisation with the MCC vaccine was performed in all regions, although the targeted age groups differed. The decision process of the health authorities was not supported by an economic evaluation (Cano Portero R, personal communication).

4.6 Switzerland

In the autumn of 2001, the Swiss government decided against a general vaccination of children and adolescents. This decision was mostly based on epidemiological (e.g. incidence of SCMD appeared to be decreasing), immunological (e.g. at that time interactions with inactivated polio vaccine could not be excluded) and biological (capsular switch risk) evidence and no economic reasons were given.[84] Subsequently, a model to evaluate the cost utility of meningococcal vaccination has been developed,[62] which is currently used to support decision making in the Swiss government (Jaccard Ruedin H, personal communication).

4.7 UK

Following the initial increases in SCMD, the Department of Health funded a comprehensive clinical trial programme concerned with MCC vaccines in 1995. The clinical trial programme was based on immunogenicity and safety studies and not on formal efficacy trials.[45,85] The introduction of the MCC vaccine was originally anticipated for October 2000, but this was brought forward by 1 year because of the early completion and favourable outcome of the clinical trials and the sustained increase in the number of serogroup C cases.[45,86] An in-house economic analysis was performed by the Department of Health, but these results were not made public. Further analyses, including cost-effectiveness studies and mathematical modelling studies,[66] that would normally be commissioned and considered before a vaccine is introduced[87] were performed after the introduction of the vaccine and have confirmed the public health and economic attractiveness of the MCC campaign. Initially, the catch-up campaign targeted individuals aged <18 years. This was later extended to persons aged <25 years as the incidence of disease continued to rise in 20- to 24-year-olds after the introduction of the original vaccine programme. The role of economic analyses in informing these changes was minimal, as public health considerations predominated.

4.8 Summary

Overall, our results show a strong positive correlation between the reported incidence of SCMD and the implementation of an MCC vaccination programme and the conduction of an economic evaluation. Of the ten countries that had implemented an MCC vaccination campaign by 2003, eight were among the ten countries with the highest SCMD incidence (table I). Economic evaluations were identified for six countries that all belonged to the ten countries with the highest SCMD incidence (tables I and V).

Table VI shows typical health economic questions decision makers ask when having to decide about a new vaccination programme. In order to address these questions quickly, highly flexible models are needed that can be adjusted easily.

Table VI

Typical questions decision makers wished to see resolved in relation to a new vaccination programme^a

5. Discussion

Vaccination campaigns seem to be especially promising candidates for economic evaluations; they have a large budget impact because of their size and annual recurrence and they are usually planned and implemented by the government. Of the 21 countries we looked at, 11 did not engage in MCC vaccination, but these decisions do not seem to have been supported or influenced by economic evaluation. Of the other 10 countries, six implemented vaccination programmes without prior economic evaluation. Thus, we found only four countries (Australia, Canada [Quebec], The Netherlands and the UK) where economic evaluations were performed prior to decision making on MCC vaccination. In Portugal, Switzerland and the UK, further economic evaluations that have either already played (Portugal and Switzerland) or may (UK) play a role in the reassessment of the current strategy are now available.

Our results suggest that a high or increasing incidence of SCMD (table I) and the associated anxiety in the population were the key factors for considering and implementing MCC vaccination campaigns, as well as for conducting economic evaluations of such strategies. Economic evaluations were utilised particularly to identify the most efficient vaccination strategies. Economic evaluations appeared to be more influential on the type of routine childhood vaccination schedule than on strategies for mass vaccination campaigns. Australia, Quebec (Canada) and The Netherlands all added the MCC vaccine to their childhood vaccination programme at age 12 or 14 months, instead of at ages 2, 3 and 4 months, as this was the most cost-effective strategy. For mass vaccination, Australia and Quebec (Canada) chose to implement the most effective mass vaccination campaigns, which were less cost effective than the alternatives (see sections 4.1 and 4.2).

Beutels et al.[88] argued, based on a literature search, that the frequency of economic research on vaccine preventable diseases is associated with the existence of multiple target groups (e.g. for hepatitis A virus, travellers, healthcare workers, children and military personnel) and multiple vaccines (e.g. polysaccharide and conjugate vaccines), demanding separate analyses in different settings. They concluded that economic evaluation is not yet a standard ingredient of policy making and that it would more likely be employed to add support for programmes that lack popular appeal (e.g. hepatitis B and varicella zoster virus vaccination).[88] In countries where SCMD incidence had been rising dramatically, vaccination had no shortage of public support, and decisions had to be made quickly, which could explain the paucity of studies prior to decision making on this subject.

Economic evaluation has been used to support decisions on various vaccines (particularly regarding pneumococcal conjugate vaccine[89] and hepatitis B virus vaccine) in many industrialised countries[88] and it may be used more consistently for future decisions. The WHO[90] and the National Institutes of Health[91] encourage the economic assessment of new vaccines in the developing world. The cost effectiveness of (hypothetical) future vaccines such as vaccines against cytomegalovirus, Chlamydia trachomatis and human papillomavirus, has also been studied, e.g. by the Institute of Medicine.[92] Promising future vaccine candidates for the Dutch vaccination programme have been selected on the basis of burden of disease and cost-effectiveness data.[93] A recent US study showed the high importance for academic, government and industry groups involved in US vaccine policy of economic evaluations of new vaccines.[77] For the American Centers for Disease Control and Prevention and the American Academy of Pediatrics, the cost effectiveness of a vaccine is an important consideration when recommending new vaccines for routine immunisation.[6]

We consider economic evaluations to be a valuable tool to support decision making, as shown for MCC vaccination in Australia, Canada (Quebec), The Netherlands, Portugal and Switzerland. For example, Dutch decision makers at the Ministry of Health, Welfare and Sports, the Ministry of Economic Affairs and the Health Council not only greatly appreciated the availability of economic evaluation results prior to decision making, but also worked together closely with health economists to adapt the model to their information needs (see table VI).

The main disadvantages of economic evaluations are the time and resources required to conduct them. Depending on the public health problem, immediate action might be necessary, especially if there is a high level of public anxiety. In addition, often neither the money nor the experts are available to conduct an economic evaluation within the desired (short) timeframe. Furthermore, the decision makers need to have a basic understanding of economic evaluation in order to utilise the results appropriately. Nonetheless, the costs of an economic evaluation might easily be offset by savings generated from choosing a more efficient strategy.

All economic evaluations of our analysis were modelling studies (tables II and III). The issue of modelling in health technology assessment has been addressed by several publications providing overviews and general guidance.[94–96] Models are generally required in order to evaluate the potential impact and cost effectiveness of vaccination programmes. This is because clinical trials are, by necessity, small, and vaccination programmes are usually applied to large proportions of the population. Furthermore, immunisation programmes have long-term effects that may not be measurable over the course of a clinical trial. Thus, there is a need to extrapolate to the population level and over a long period of time. Modelling is also necessary to determine the effectiveness on a population level and to assess the uncertainty by sensitivity analysis.

Immunising against an infectious disease may confer additional benefits to others in the population, i.e. following a vaccination campaign, the risk of disease declines even in unimmunised individuals because there are fewer infectious individuals to whom they may be exposed. Studies in the UK have shown that an MCC immunisation campaign has generated substantial ‘herd immunity’.[50] To capture these herd immunity effects, a transmission dynamic model is required. These dynamic models simulate the spread of an infection in a population and measure the impact of an intervention on transmission and disease risk.[88,97–99] The most commonly used static models derived from the decision-analysis field are not best suited for the evaluation of most immunisation programmes as they cannot take into account these indirect effects appropriately. A transmission dynamic model has been designed for N. meningitidis that offers the opportunity to predict the herd immunity effect much better than static decision models.[68] Compared with static models, dynamic models have two main disadvantages: they request detailed data about the disease transmission and show a higher level of complexity. However, the development and use of such dynamic models should be encouraged to ensure that important indirect effects are not ignored.

Clearly, modelling also has its shortcomings. Any model is only as good as its inputs, and models are only a simplification of reality — there is no perfect model. Models could also be biased towards producing the desired results, especially if the model structures and inputs are not well described. Even a fully probabilistic sensitivity analysis will be meaningless and render incorrect certainty intervals if the wrong model is chosen. Still, a good sensitivity analysis is one of the cornerstones of any economic modelling study. The sensitivity of several methodological characteristics (table IV) and their strong variation between studies (tables II and III) demonstrates the need for firmer national and international modelling guidelines and better adherence to these guidelines when conducting economic evaluations, particularly within the vaccine field. To decrease the threat of manipulation, models should always be transparent and validated internally and externally. Model validation should be performed by a specialist as it requires specific expertise, which is generally lacking amongst decision makers, public health professionals, clinicians and economists.[100] The UK’s National Institute for Health and Clinical Excellence (NICE), as well as the Dutch Committee of Pharmaceutical Help of the Board for Health Care Insurance, have specific modelling experts that check the models submitted by the industry.

6. Conclusion

Economic evaluations are a valuable tool for supporting decision making about MCC vaccination in several countries. In most countries for which economic evaluations of MCC vaccination were available, the results clearly had an important role in the decision-making process, especially in the choice of routine MCC vaccination strategy. However, most countries we contacted had not performed an economic evaluation of MCC vaccination to aid the decision-making process, either because public health considerations and public anxiety took precedence or because disease incidence was too low for a vaccine campaign to be considered.

Firmer national and international modelling guidelines and better adherence to such guidelines could help to improve the international comparability of economic evaluation results for MCC vaccination.