Evidence and policymaking: The introduction of MMR vaccine in the Netherlands.

Based on a case-study of the introduction of measles-mumps-rubella (MMR) vaccine in the Netherlands two decades ago, using documentary and archival sources, this paper examines the way evidence is used in policymaking. Starting from the question of 'what counts as evidence', two central claims are developed. First, the decision to introduce MMR was not one but a series of decisions going back at least seven years, over the course of which the significance attached to various forms of evidence changed. Second, results of international studies were coming gradually to be of greater significance than evidence gathered from within the Netherlands itself. These developments had, and continue to have, major consequences for national scientific competences. Copyright 2010 Elsevier Ltd. All rights reserved.

Introduction

In the last few years the claim that health policy, like medical practice, should be evidence based has attracted widespread support (e.g. MacIntyre, Chalmers, Horton, & Smith, 2001; Niessen, Grijseels, & Rutten, 2000). Immunisation is one area of health policy in which this claim may have particular appeal, given the challenges with which it now confronts policymakers. More and more new vaccines are becoming available, and some of them (initially at least) are very expensive. Faced with deciding which should be introduced into national immunisation programmes, a standard set of criteria – disease burden, the safety and effectiveness of the vaccine, the cost-effectiveness of vaccination and so on – has obvious appeal (Gezondheidsraad, 2007; Smith, Snider, & Pickering, 2009; Vyse et al., 2002). Application of such criteria depends on reliable epidemiological, scientific and economic data being available. In many societies growing scepticism confronts health policymakers with a very different challenge. Because it affects whole populations, and their most vulnerable members in particular, immunisation has always been subject to public scrutiny (e.g. Colgrove, 2006). But today it appears that more and more people are questioning vaccine safety or of the benefit of vaccination. A decade of controversy over the safety of measles mumps and rubella (MMR) vaccine; a far less than enthusiastic response to the introduction of cervical cancer vaccine (HPV) in many places; the aftermath of recent pandemic influenza scares: all point to an erosion of public confidence in immunisation programmes. Policies need to be justified publicly to an extent that was not previously the case. Here too, evidence for risks to public health, robust enough to assuage doubts, should help ensure the transparency and the legitimacy of decisions taken. Good evidence, in other words, can support immunisation policymaking in a number of ways. Empirical research suggests that in practice the relations between evidence and policymaking are varied and complex.

Whilst some studies find decisions to have been evidence based (e.g. Wenger, DiFabio, Landaverde, Levine, & Gaafar, 2000), others show policymakers acting despite a lack of epidemiological data and acknowledged uncertainties regarding the efficacy of a new vaccine (e.g. Haas et al., 2009). Public perceptions of the risks of a disease, that do not necessarily correspond to actual risk, may create political pressure to introduce a vaccine even where data are lacking (Levine & Levine, 1997). In yet other studies the possibility of local production of a vaccine appears to have been an important consideration in the decision to introduce it (DeRoeck, Clemens, Nyamete, & Mahoney, 2005; Milstien, Kaddar, & Klenny, 2006; Munira & Fritzen, 2007). Many of these studies suggest that a variety of actors with distinctive interests and values are typically involved in policymaking. Though Ministries of Health and their advisory committees generally play a central role, the organized medical profession, national and international vaccine manufacturers, and consumer advocacy groups may all actively seek to influence decision-making (Munira & Fritzen, 2007; Wenger et al., 2000). In such debates, which differ from country to country, participants deploy evidence that seems to support their particular interest or perspective (Haas et al., 2009). A study of the introduction of hepatitis B vaccine suggests that decision-making may differ between early and late adopters: that scientific evidence, and assessments of political and technical risks involved in vaccine adoption, may differ from country to country. For countries contemplating early adoption of a new vaccine, for whatever reason, lack of evidence may be an important uncertainty. On the other hand for late adopters formal recommendations by agencies such as the WHO may weigh heavily in decision-making (Munira & Fritzen, 2007).

What influences the way in which evidence in practice figures in the decision to introduce a new vaccine? This question provides the starting point for the present paper, that focuses on the introduction of measles mumps and rubella (MMR) vaccine in the Netherlands two decades ago. It has been suggested, in regard to Evidence-Based Medicine, that ‘what counts as evidence’ is far from straightforward (e.g. Lambert, 2006). In arguing that the same holds for evidence-based health policy, this paper makes two central claims. First, we try to establish that the decision to introduce MMR in the Netherlands was the outcome of a protracted process in the course of which the significance of different kinds of evidence changed. Second, we argue that this change shows a growing reliance on international data at the expense of national data, and that this development has major consequences for national scientific competences. This argument is consonant with a conclusion other authors have reached with respect to evidence-based health policymaking more generally (Behague, Tawiah, Rosato, Some, & Morrison, 2009).

Method and analytical framework

Most published studies of vaccine introduction, focusing on decisions recently taken or in course of being taken, rely principally on interviews with policymakers. Historical research, focusing on decisions previously taken, provides a different perspective on policymakers’ concerns and uncertainties. It permits us to explore not only changes in policy over time, but also influences that policymakers may not have cared to voice, and the ways in which differences were ultimately resolved in practice (Muraskin, 1998; Stanton, 1994). In addition to the scientific literature and a review of articles appearing in the principal Dutch medical journal, in carrying out this study we have consulted the Annual Reports (Berichten) of the State Institute of Public Health (Rijksinstituut voor Volksgezondheid) from 1960 to 1987; and the reports, minutes of meetings and correspondence of the Health Council of the Netherlands (Gezondheidsraad), its Committee on Immunisation (Commissie Beraadsgroep Immunisatie) and the sub-committee on rubella and mumps in which the decision to introduce the vaccine was prepared.

The analysis makes use of the conceptual scheme proposed by Dobrow et al. for analysing the role of evidence in health policymaking (Dobrow, Goel, & Upshur, 2004). These authors contrast the conventional view, according to which ‘what constitutes evidence’ is unambiguous and subject to agreed standards of reliability and validity, with what they term a ‘practical-operational’ view. From this second perspective, what in practice is taken to be evidence is influenced by the context in which decisions are taken. A conceptual scheme is developed, by means of which the question of how this is so can be addressed. The scheme involves two sets of distinctions: between internal and external contexts, and between introduction, interpretation, and application of evidence. The ‘internal context’ of decision-making is a matter of purpose (what precisely is to be achieved?), of participation (what perspectives and interests are involved?), and of the decision-making process. The ‘external context’ includes the geographic, demographic and epidemiological characteristics of the disease, as well as a range of social, economic and legal issues (including the capacity and organization of health services, and the political interests that need to be taken into account). The relevance of these two contexts is to be analysed in relation to distinctive stages of a decision-making process. What evidence is introduced into decision-making, for example, depends on who participates (and thus on whether or not conflicts of interest arise). It may also be influenced by external contextual factors (such as the distribution of a disease in the population). Interpretation of evidence will depend on who is doing the interpreting, and on the time and effort available for analysis. Important here is whether or not evidence gathered elsewhere is seen as applicable to the specific context for which a decision is being made. Finally in the application of the evidence, the external context is of great significance. Is a decision that seems to follow from the previous two stages ideologically acceptable and politically and economically feasible? This conceptual framework leads to a refinement of the first claim we will try to develop: changes in ‘what counted as evidence’ through the course of MMR decision-making should be viewed in the light of changes in the context in which decision-making took place.

Background: three diseases and three vaccines

In the 1960s scientists set about developing vaccines against measles, mumps and rubella. Though it was viewed as the most serious of the childhood infections, measles in the industrialized world was no longer the killer disease that it had once been, and that in poor countries it still is. Mortality rates (largely due to secondary infections) had been greatly reduced by the use of antibiotics. However, incidence of the disease was not falling, and because the disease was so widespread absolute numbers were significant and the burden on parents and on physicians considerable. Vaccination against measles was widely viewed as highly desirable, even though some feared that protection in childhood might result in infection being postponed to later in life, when its effects could be more serious. In 1968 measles vaccination began in England and Wales, and despite an initial uptake of only 50% incidence fell dramatically.

Mumps was seen as a far less serious disease than measles. In the 1960s the common view among European medical professionals was that though unpleasant, childhood mumps was a mild disease (Wilterdink, 1984). The view in the United States may have been different (Offit, 2007, 21, 29–30). Potential complications include orchitis (an infection of the testes) in males, and aseptic meningitis (or mumps encephalitis), but these were rare and could generally be treated successfully. Widespread popular belief notwithstanding, male sterility very rarely followed mumps infection. British and European physicians in the 1960s and 1970s did not agree as to whether mumps vaccination was needed, referring to its unserious, though painful nature, and the risk of mass immunisation of children leading to increased incidence among adults, with potentially more serious consequences (British Medical Journal, 1980; Lancet, 1974).

Rubella (German measles) provoked no such doubts. In 1940 it had been established that rubella contracted during pregnancy could lead to spontaneous abortion, to central nervous system defects, or to one of a range of other serious and debilitating conditions in a new-born child (congenital rubella syndrome, or c.r.s.). Rubella was one of the few known viral causes of congenital malformation. The epidemic that struck the USA in 1963–1965 was particularly serious, and was said to be responsible for some 20,000 foetal deaths and a similar number of brain-damaged children. Public and political pressure to produce a vaccine built up (Galambos & Sewell, 1995, 105).

When the first rubella vaccine was licensed, in 1969, the question was not whether it was needed, but to whom it should be given. In the USA it was decided that all girls and boys aged between one year and puberty should be given the vaccine, thus blocking spread of the virus among children. The reasoning was that if herd immunity in the group among which virus circulated most particularly could be achieved, pregnant women would be protected indirectly. Vaccinating young women of child-bearing age directly with a live virus vaccine was considered too dangerous. By contrast, in Europe not all were convinced that this strategy was the most appropriate. Thus when rubella vaccine was licensed in Britain, in 1970, a large scale vaccination programme directed initially at girls age 11–13 was favoured. Eradication of the disease, by tackling the reservoir of infection in children of both sexes, could be a later objective.

Medical (and public) opinion regarding the need for a vaccine differed significantly as between measles, mumps and rubella. There were also important uncertainties: regarding likely uptake and duration of protection (and so the risk of a shift in incidence to older age groups where infection could be more serious); regarding possible side effects; and (especially as far as rubella was concerned) regarding the most appropriate vaccination strategy. These uncertainties were compounded by the emergence of competing vaccines with different properties.

Measles, mumps and rubella are viral diseases, and vaccine development made use of the tissue culture techniques pioneered by John Enders and his Harvard colleagues in the late 1940s. As with polio vaccine development a few years previously, two approaches were available. One was to inactivate the virus, so producing a ‘killed’ virus vaccine. Technically simpler, this was how Jonas Salk had made his polio vaccine. The other was to weaken (or ‘attenuate’) the virus by passing it one or more times through an appropriate cell culture: the strategy that Albert Sabin had followed. Investigators were faced with a number of difficult judgements: whether to try to produce an inactivated or an attenuated virus vaccine; how to inactivate or in what medium to culture the virus; how much attenuation was required in order to produce a vaccine that was both safe and effective. In each case the starting point was a locally available strain of the virus, and these differed from one centre to the other. Different clinical isolates of a given virus can differ genetically, leading to differences in immunogenicity and toxicity. Difficulty was compounded by the lack of a suitable animal model. Monkeys, which had been widely used in testing the safety and potency of polio vaccines, proved unsuitable for testing these vaccines. It was difficult to determine at what point in the attenuation process to start testing a vaccine on humans.

A measles vaccine came first.1 In 1960 John Enders isolated a measles strain from a boy named Edmonston and attenuated it using a culture of chick embryos. In the years following numerous vaccines were developed, many of them starting from Enders’ Edmonston isolate. In early 1963 the first two measles vaccines were licensed in the USA: a live vaccine containing the Edmonston B strain produced by Merck (Rubeovax), and a killed (inactivated) vaccine produced by Pfizer. Others followed. Studies carried out under the auspices of the World Health Organization (WHO) in eight countries found great differences between 11 live virus vaccines tested. The WHO team also concluded that the inactivated vaccines were ineffective and should not be used, though some felt that such a judgement was premature (Norrby, Lagercranz, & Gard, 1965). It then appeared that inactivated vaccine led to “atypical measles” in subjects later challenged by the virus, and it was withdrawn from the American market. By the late 1960s, a number of live measles vaccines were in widespread use. Merck had further attenuated Edmonston B vaccine. Now using the so-called ‘Moraten’ (‘More Attenuated Enders’) strain it was renamed Attenuvax. Other manufacturers, including Glaxo and Burroughs Wellcome, used other strains.

Whilst development of a vaccine against mumps might have lacked the urgency that the measles vaccine had had, nevertheless, various centres were working on them in the 1960s. Here too, killed virus vaccines were soon abandoned. Live virus vaccines were produced starting from local isolates and attenuating them in different ways. Widely used ones included the Urabe (developed in Japan and then used by a number of European manufacturers), Leningrad-3 (produced in the then Soviet Union), and L-Zagreb (developed in then Yugoslavia by further attenuating the Leningrad-3) strains. Principally used in the USA was the so-called Jeryl Lynn strain, named after Maurice Hilleman’s daughter Jeryl Lynn, from whom the virus sample was initially taken. Trials showed that the vaccine was safe and effective in a single dose: there seemed to be no side effects and immunity persisted for a number of months (Offit, 2007, 30). Merk’s Mumpsvax was licensed in the USA in 1967.

In the case of rubella too, the search for a killed virus vaccine was soon abandoned. Here too competing vaccines emerged, based on strains obtained in different places and attenuated in different ways. The so-called High Passage Virus (HPV-77) strain was developed by Paul Parkman (at the National Institutes of Health’s Division of Biologics Standards) by passing a virus isolate 77 times through a culture of African green monkey kidney cells. At Merck, Hilleman isolated a virus strain from a Philadelphia schoolboy named Benoit in 1962. As noted earlier, in the mid 1960s the United States was struck by a severe rubella epidemic and pressure to have a vaccine available rapidly built up. Worried that disagreement over which was better could slow things down, and believing that licensing would be more rapid if the NIH vaccine were chosen, the influential medical philanthropist Mary Lasker persuaded Hilleman to abandon his vaccine in favour of Parkman’s (Offit, 2007, 74). Huygelen and Peetermans at RIT in Belgium – later taken over by SmithKline & French – were working with a rubella strain called Cendehill, cultured first on monkey and later on rabbit kidney cells (Parkman, 1999). At the Wistar Institute in Philadelphia Stanley Plotkin chose to grow his virus in a culture made from aborted foetuses (Plotkin, 1996, 276). The important point for Plotkin was that they would be free of the contaminating pathogens that had been found in many animal cell cultures. The vaccine they developed was called RA27/3. Between 1967 and 1969 clinical trials of a further attenuated RA27/3 vaccine were conducted in various countries. However many scientists, including Albert Sabin, feared that using human cells to culture viruses carried its own risks (Offit, 2007, 86). On moral grounds, too, there was resistance to the idea of growing vaccine in foetal cells, particularly in the USA. Thus whilst a number of European manufacturers started working with RA27/3 (including Burroughs Wellcome in the UK and Institut Mérieux in France) initially their American competitors would not.

By 1970 a number of rubella vaccines were available. In the USA the Merck, Philips-Roxane and SmithKline & French vaccines were all licensed early in the year. In the United Kingdom SmithKline’s Cendevax and Almevax (the RA27/3 vaccine manufactured by Burroughs Wellcome) were introduced later the same year. Discussion of the relative merits of the different vaccines was more protracted and more complex than was the case with measles or mumps vaccines. This reflected the fact that control of rubella presented a unique problem, “ in that the ultimate goal of immunisation is a remote one, namely the protection of some future fetus against damage from intrauterine infection” (Horstmann, 1975). A major concern was the possibility of immunity declining with time, since this could result in a more susceptible population of women of child-bearing age (Parkman, 1999). Despite studies showing little difference between the vaccines, or complex and varying changes in immunity over time, consensus in favour of the RA27/3 vaccines emerged, on the grounds that they provided longer lasting immunity. In 1978 the decision was taken at Merck to replace the HPV-77 strain with the RA27/3 strain, and this was finally licensed in the United States in January 1979.

Discussion of the relative merits of the different measles, mumps and rubella vaccines was gradually transformed by the emergence of combined vaccines. The major argument for combining different vaccines was, and is, that schedules can be simplified, and so uptake increased. At Merck Hilleman had begun studying antibody responses to various vaccine combinations, including the three component measles mumps and rubella, in 1969 (Hilleman, 1969). Their MMR vaccine was licensed in 1971, alongside the individual vaccines. Merck soon became the only domestic supplier of MMR vaccine in the USA. Since low coverage rates had long been a matter of concern in the USA the combination vaccine had particular advantages, and MMR began to be used routinely by the mid 1970s. In Western Europe coverage rates were higher, and it was only in the course of the 1980s that European countries switched to MMR. Sweden, one of the first, introduced it in 1982, the UK in 1988. In the Netherlands vaccination with MMR began in 1987 but, as we shall now show, the process culminating in its introduction began many years earlier.

Case study: vaccination against measles mumps and rubella in the Netherlands

Ever since the establishment of the Dutch national immunisation programme (RVP), in 1957, the State Institute of Public Health (Rijksinstituut voor Volksgezondheid or RIV – from 1984 RIVM) had responsibilities not only for surveillance of public health but also for providing the vaccines the immunisation programme required. It could do this either by developing and producing them itself or, where necessary, by purchasing them from a commercial manufacturer. Sustaining both the country’s highly effective immunisation programme as well as its own scientific and technical competences was central to the Institute’s work, with transfer of vaccine technology to developing countries as an important additional objective. Its achievements, especially in regard to polio vaccine, were widely acknowledged (Blume, 2005). Provision of a suitable vaccine, and ensuring its safety and efficacy, was the responsibility of the RIV, whilst the Health Council (Gezondheidsraad) had (and has) responsibility for strategic oversight of the national immunisation programme, and for advising the Minister of Health on possible changes.

Rubella and measles: commitments and controversies

Vaccination against rubella and measles, which began in the Netherlands in 1974 and 1976 respectively, followed a decade of work by the RIV. In 1964 the RIV had begun studying both diseases, with a view to producing vaccines for use in the national immunisation programme. Their general objectives, and their recent success with polio vaccine, shaped the way in which they set about doing so. By incorporating (inactivated) polio vaccine in the combined diphtheria pertussis tetanus (DPT) vaccine previously in use, it had been possible to add it to the vaccination schedule without the need for potentially disruptive changes. The hope was that this strategy could be repeated, by adding an inactivated measles vaccine to the existing DPT-P. Unconvinced by reports of potential risks involved in use of inactivated measles virus, the RIV succeeded in producing a five component DPTP-M in 1969. In 1970 this was tested in small-scale trials. However, not all Dutch experts agreed that an inactivated vaccine should be used. Influenced by research abroad, some were starting to favour live virus vaccine. Nor were politicians willing to wait until the RIV had produced its vaccine. In 1976 measles vaccination began using Attenuvax, purchased from Merck. The RIV then set about preparing its own live measles vaccine, using the ‘Moraten’ strain obtained under license from Merck. By 1981 the RIV’s own live measles vaccine was in use (However, work on an inactivated vaccine, in collaboration with Swedish colleagues, continued throughout the 1980s. RIV researchers believed that problems reported with earlier inactivated vaccines could be overcome, and that a technology particularly suited to conditions in developing countries could be based on the inactivated vaccine). Meanwhile measles vaccination in the Netherlands had been a striking success, with cases reported falling from more than 2000 in 1976 to less than a hundred per annum by the early 1980s.

Production of a rubella vaccine confronted RIV with a different set of technical issues. By 1971 an HPV77 strain obtained in the USA had been further attenuated, giving a vaccine known as HPV77-RK5. Trials conducted in 1971 and 1972 showed that this appeared to be comparable with commercial vaccines in terms of effectiveness and safety. By 1973 the RIV had concluded that the HPV77-RK5 vaccine should be added to the national immunisation programme, and in 1974 it was.

Whilst it was immediately clear that measles vaccine should be given to all children in the first year of life, it was unclear how the rubella vaccine should be used. In 1969 the Health Council was asked to advise on the desirability of rubella vaccination, and on the strategy to be followed. Considering the strategy adopted in the USA (aiming at protection of the community by reducing, and ultimately eliminating, virus circulation) and the alternative strategy adopted in the UK and elsewhere in Europe (aiming at protecting individual women who might later become pregnant) the Council concluded that there was insufficient evidence for the former strategy. Until such evidence became available the second strategy should be followed (Gezondheidsraad, 1971). Thus, from 1974 11 year old girls (only) were vaccinated. Reported cases fell from 2 to 3000 per year in the early 1970s to 7–800 in the years thereafter – though with occasional peaks. However a crucial uncertainty remained. Would protection conferred on young girls persist till the time of an eventual pregnancy?

In 1979 the Chairman of the Health Council established a consultative body on immunisation (CBI), which soon turned its attention to rubella vaccination. By early 1982 there was a clearly expressed wish to look at the matter in the light of experience abroad (particularly in the UK and Sweden). This explicit attention to experience abroad led the Committee to reconsider both the purpose of rubella vaccination and the way in which evidence was being used. Whilst vaccinating young girls only had succeeded in reducing the incidence of c.r.s. and rubella-related pregnancy terminations, British experts doubted that it could lead to elimination of c.r.s. (Miller, 2007). Also at this time mathematical modelling of infectious diseases, though still in its infancy, was acquiring growing influence. A British simulation modelling study suggested that the preferable strategy depended on the efficacy of the vaccine, on the percentage vaccinated, on the rate of decay in immunity over time, and on the relative importance given to short term benefits as against the risk of subsequent resurgence (Knox, 1980). The CBI wished to explore the utility of the different strategies through simulation modelling using data from the Netherlands. When the Minister of Health invited the Council to “critically consider the place of rubella vaccination in the National Immunisation Programme,” a special committee was installed to prepare the advice to the Minister. In the event, it was decided that this committee would respond both to this request and a second, regarding mumps, discussed below.

Minutes of the meetings that took place in 1983 show the new committee becoming convinced that the rubella vaccination strategy should be changed. A number of arguments played a role. Research elsewhere suggested that the American approach, or a combined one involving a second shot for 11 or 12-year-old girls, seemed to be best. Increasing numbers of European countries were now following the American (or the combined) strategy. Feeding available data into a simulation model suggested that the combined strategy would be best for the Netherlands, and that only this approach would lead to reduction in virus circulation (Gezondheidsraad, inventory 383). By late 1983 the committee had agreed to advise the Minister to replace the current rubella strategy by one in which both boys and girls would be vaccinated at the ages of 1 and 9 years. The expectation was that this would lead to the virtual elimination of rubella within 5–10 years (It was also estimated that rubella could well become more prevalent within the small communities opposed to vaccination on religious grounds. Committee members felt this was no reason to question a strategy deemed to be in the interest of the majority).

The committee repeatedly discussed the question of the rubella strain being used in the vaccine. RIV made use of the HPV77 strain, whereas commercial manufacturers were all switching to the RA27/3 strain, reported to have a higher immunogenicity. Responding to changing international opinion, in 1981 a study had been carried out in Rotterdam. Half of the girls in the study were given the existing RIV vaccine, and half were given the RA27/3 vaccine Almevax. RIV researchers concluded from this study that both vaccines produced a more than adequate response. They saw no reason to switch to an RA27/3 vaccine (RIVM , 1981). The Institute’s representative in the committee continued to argue this point of view: both vaccines were perfectly good, and any differences in benefit minimal. A switch would be expensive, and would yield virtually no benefit. Another committee member, a Belgian virologist, argued that the Netherlands should switch, as Belgium had done, for reasons that had most to do with public confidence. Whatever studies showed, there was a risk of controversy and public loss of trust if people believed they were being given an inferior vaccine (Gezondheidsraad, inventory 387). Differences of opinion could not be resolved, and the committee made no recommendation regarding preferred vaccine. In 1984 the Institute started working with the RA27/3 strain, cultured on human diploid cells in place of rabbit kidney (RIVM , 1984).

Introduction of MMR

In the Netherlands, as elsewhere, in the 1960s and 1970s mumps was seen as a minor disease of childhood. It received little attention in the Dutch medical press. In 1971 Merck’s European subsidiary, MSD, applied for permission to import Mumpsvax to the Netherlands. The government’s view, based on advice from the Health Council, was that whilst there was no reason to deny permission to import the vaccine, nor was there reason to consider mass vaccination against mumps at that time. Seeing no imminent need, the RIV had devoted no attention to development of a mumps vaccine.

When the Health Council’s CBI committee began to consider the possibility of mumps vaccination, in 1980, disagreement emerged regarding what evidence was most relevant (Gezondheidsraad, inventory 1509). One member, a microbiologist, attached particular weight to what was known of the disease. Mumps in children was relatively innocuous, and mumps-related mortality was minimal. Moreover, little was known of the duration of protection that vaccination would provide. There was the risk of circulation being shifted to older age groups. The Belgian virologist pointed to experience abroad. The vaccine was already being used in the USA, Germany and Sweden. A representative of the government Health Inspectorate noted the likelihood that the European Parliament would recommend harmonisation of vaccination schedules in Europe: mumps vaccination was a principal difference between schedules in different European countries. Economic arguments were introduced. They suggested that the cost of providing the vaccine would be less than costs associated with the hospital admission of mumps patients. Other things being equal, money could be saved.

Though mumps-related mortality was minimal many general practitioners were offering mumps vaccination to parents and import of mumps vaccine was rising. Crucially, economic considerations were becoming more significant in the committee’s deliberations. Looking at the numbers admitted to hospital with mumps, and the lengths of stay, the Health Inspectorate had estimated that in 1979 these hospital costs had been twice what had been spent on measles vaccine (Gezondheidsraad, inventory 1510). When the Minister formally requested the Health Council to advise on the desirability of mumps vaccination, it was agreed that the committee looking into rubella would consider mumps too. When it met for the first time in April 1983 it appeared that one of the arguments for considering mumps vaccination had lost some of its force. The number of doses imported annually had fallen for two years in succession. Nor was the seriousness of the condition a major concern, since prevailing medical opinion was still that serious complications were rare. The duration of protection provided and the consequent risk of enhanced virus circulation among older age groups remained a source of uncertainty. Available data were inadequate for mathematical modelling to be of much value. Two arguments in favour of mumps vaccination gradually came to dominate the committee’s deliberations. One referred to practice elsewhere. Dutch experts were impressed by data from the USA showing an enormous fall in the number of reported cases of mumps. At the same time it was increasingly accepted that Dutch vaccination practice should be harmonised with those of other European countries. The second important argument related to costs and benefits. Although data for much more than a back-of-the-envelop calculation were not available, potential savings to the health care budget figured largely in the committee’s discussions. By September 1983, the broad lines of the recommendations were becoming clear. All children should be vaccinated against mumps at the age of 14 months. A second shot should follow later. This would help deal with the risk of immunity declining over time, but it also reflected what by now was becoming the virtual certainty that the Netherlands would switch to a combined MMR vaccine.

The committee’s report also considered whether mumps vaccination should be limited to boys, given that boys seemed to be more susceptible and that adult men faced particular risks. Since vaccinating boys only was incompatible with use of the combined vaccine, that would have to be given to all children, arguments were needed for vaccinating girls too. One was that a ‘significant’ number of girls were also admitted to hospital (an average of 115 per annum). A second was that vaccinating only half the children would insufficiently reduce virus circulation, so that passive protection of the non-immunised would be limited.

Whilst these discussions were ongoing, the Ministry of Health requested RIV to consider what the options would be, were the Health Council to advise in favour of mumps vaccination. Acknowledging that this was likely, and that it had no expertise in the mumps field, in 1981 the RIV had begun considering the options. The initial intention was to produce a mumps vaccine which could then be mixed with their existing measles and rubella vaccines. A mumps strain would have to be obtained from a commercial manufacturer. Given doubts that had been raised regarding the safety of Urabe, the preference was for the Jeryl Lynn strain. Negotiations with Merck’s European subsidiary (MSD) regarding the production of mumps vaccine under license began in 1982, but broke down when it became clear the company would consider licensing its combined MMR vaccine, but not the single mumps component. Anticipating the forthcoming change in the immunisation programme, RIV (that in 1984 was reorganised as RIVM), continued negotiations with this and two other manufacturers regarding production of MMR under licence.

In June 1984 the Health Council advised that MMR be introduced to the Dutch immunisation programme. Financial considerations weighed heavily, both with the Health Council and with the Ministry of Health. In September 1985 the government Minister responsible requested the Health Insurance Council (Ziekenfondsraad) to consider the financial implications of introducing MMR. Despite the fact that no cost-benefit data regarding MMR were available for the Netherlands, the Minister pointed out that “data from other countries, notably the United States, Switzerland and Austria show that MMR is highly cost-effective. It seems reasonable to conclude that the benefits of MMR vaccination does more than balance the costs” (van der Reijden, 1985).The new arrangements would take effect in January 1987. MMR vaccine would be given to all children at the ages of 14 months and 9 years.

Discussion: evidence in context

The first claim that we wished to substantiate was that in course of a decision-making process the significance attached to different forms of evidence may change, as a result of changes in the context in which decision-making takes place.

The decision-making process that resulted in the addition of MMR vaccine to the Dutch national immunisation programme in 1987 can be said to have started in 1980. It was in that year that the question of mumps vaccination arose in the Health Council’s Committee on Immunisation (CBI). Since immunisation of all children against measles, and 11 year old girls against rubella had already much reduced the incidence of these diseases, and since mumps was viewed as clinically unimportant, there were scant epidemiological grounds for changing the programme. The RIV, an important participant in vaccine policy discussions, had devoted years of effort to developing and producing the vaccines against both measles and rubella then in use. Having seen mumps as unimportant it had not invested in development of a mumps vaccine, and had no institutional interest in pressing for mumps vaccination.

A series of shifts in both the internal and the external contexts of decision-making then began to undermine this status quo. First, opinion internationally was converging on the view that the purpose of vaccination against rubella should no longer be protection of individual women before they reached childbearing age, but rather reduction of virus circulation in the population as a whole. Reflecting this change of view other European countries, including the UK, were starting to vaccinate all children against rubella, and Dutch experts showed growing interest in their experiences. The increasingly international orientation led to the introduction of a new form of evidence in Dutch policy deliberation. Simulation modelling suggested that circulation of rubella virus could best be brought about by vaccinating all children. At the same time, the RIV’s use of the HPV-77 rubella strain was being called into question. Local studies had shown no difference in degree of protection between their own vaccine and the RA27/3 strain being used increasingly by commercial manufacturers. In the RIV’s view a change would be costly and offer little if any benefit. However in many countries the early 1980s were marked by growing public concern with vaccine safety, relating in particular to pertussis vaccine (Blume & Zanders, 2006). Maintaining public trust was becoming an important consideration, and one committee member argued that this meant that the vaccine generally perceived as better had to be adopted, if data were ambiguous. It is ironic, in view of this, that in the late 1990s controversy would come to surround MMR vaccine itself (Horton, 2004; Leach & Fairhead, 2007, chap. 4).

When discussion of the desirability of immunising against mumps began, first in the CBI and then, responding to a formal request from the Minister, in the rubella (and then also mumps) sub-committee, various kinds of evidence were introduced. They included the epidemiology of the disease (its minimal fatality and limited serious sequelae), but also practice in other European countries, as well as economic considerations. As discussion proceeded, and the implications of the various kinds of evidence debated, economic and international considerations became increasingly important. Rough calculations suggested that the costs of vaccinating against mumps would be more than outweighed by savings in medical care provided to affected children. The significance of practice elsewhere was changing: in Dobrow et al.’s terminology from ‘evidence’ to ‘context’. Whereas previously practice in other countries (e.g. the UK) had been seen as evidence in relation to choice of rubella strategy, in suggesting that immunisation schedules in Europe might have to be standardized, government advisors were in effect internationalizing the context in which decisions were legitimately to be made.

The formal structure within which expert advice to the Minister was prepared did not change between the late 1970s and the late 1980s. Nevertheless, the suggestion that European vaccination schedules were likely to be standardized can be seen as a pre-emptive shift both in the context and in the evidence deemed appropriate for decision-making in a member state of what would soon become the European Union. Moreover, within the stable advisory structure subtle changes were taking place. It seems that the expertise in vaccine development and production RIVM representatives brought with them carried less and less weight in discussions. Perhaps this too can be understood in terms of the broader ‘external context’ of decision-making. Internationally, political commitment to domestic vaccine production, where it existed, was declining in the 1980s. In a number of countries, including Australia and Sweden, it was decided that vaccine production should no longer be a task for the public sector.

Changes in the context of immunisation policymaking, and in the weighing of different forms of evidence, occurred in tandem. But in this particular area of health policy at least, there is a direction to the changes taking place. Twenty years after the events described in this paper, efforts to standardize immunisation schedules in the European Union are continuing, even though some experts question the explicit arguments used to justify such efforts (Wiese-Posselt, Reiter, Gilsdorf, & Krause, 2009).

Our second claim was that changes over time entailed a growing reliance on international rather than national data, and that this had important consequences for national scientific competences.

In 1986 the Dutch Minister of Health justified his decision to introduce MMR to the national vaccination programme by reference to studies conducted elsewhere, and despite the lack of cost-effectiveness studies carried out in the Netherlands itself. This statement points to more than the overriding importance that economic evidence had now acquired. It hints also at a shift in the relative weight coming to be attached to national as against international data. For the RIVM, with responsibilities for public health surveillance (and so for the production of national epidemiological data) and for vaccine production, the consequences were profound. The RIVM was obliged to provide the combined MMR vaccine that the Minister had decided would be introduced. How was this to be accomplished? Looking back, the then head of the RIVM’s vaccine department recalls that “The end of the story is that we had to forget about our own measles and rubella. We could only do it by producing the threefold vaccine under license from one of the companies. That’s what happened. I negotiated with the Belgians, and the French, and with Merck and we ultimately chose for Merck. It was impossible otherwise” (Ruitenberg, 1998).

Though moves toward European integration taking place in the 1980s certainly influenced the course of decision-making in the Netherlands, more general forces were, and are, at play. A recent study found health policymakers in five very different countries agreeing that policy options based on national research were no match for policies based on international research and promoted internationally. This study concluded that, “importing evidence-based policies derived in settings outside their own country undermines national experts’ experiential knowledge and the credibility of locally-generated solutions” (Behague et al., 2009, p. 1543). Policymakers in this study came from five countries in which the implementation of new programmes depended on financial support from international donors. Yet in the case of vaccine introduction, at least, something similar seems to be true for richer countries too. A study of recent decisions to introduce HPV vaccine in seven countries found that, despite major differences in the nature and timing of public debate, and despite acknowledged uncertainties regarding the vaccine’s real-world effectiveness, all seven countries introduced HPV vaccine at more or less the same time (Haas et al., 2009). The authors concluded that “those in policy decision-making roles are responding to specific, and similar, political pressures,” though without explicitly discussing what those pressures might have been. It is our contention, in the light of this study, that processes of globalization of vaccine manufacture and of health policy formulation, that began in the 1980s, are imposing a new logic on ‘what counts as evidence’ in immunisation policymaking.