Measles in France : the epidemiological impact of suboptimal immunisation coverage

La promotion de la vaccination en France depuis 1983 a abouti à une réduction de 97% de la morbidité et de 60% de la mortalité de la rougeole. Cependant, le taux de couverture stagnant et suboptimal à 84% entraine un déplacement de l’âge des cas dans des tranches où les complications sont plus fréquentes et sévères. Ainsi, la proportion des plus de 10 ans est passée de 13% en 1985 à 48% en 1997, la transmission de la rougeole se maintenant en France. Pour éliminer la maladie, une couverture vaccinale à 2 doses et à plus de 95% serait nécessaire.

The promotion of immunisation in France since 1983 has resulted in a 97% reduction in morbidity and a reduction of 60% of mortality.However, the stable and sub-optimal coverage around 84% leads to a shift in higher age groups, where complications are more frequent and serious.The proportion of those aged over 10 years was 13% in 1985 and reached 48% in 1997, the transmission of measles being maintained in France.To eliminate the disease, vaccine coverage with 2 doses and over 95% would be necessary.
T he incidence of measles has decreased in many countries around the world since the introduction of immunisation.
In 1994, a policy was initiated in the Americas to eliminate the disease.This has resulted in the interruption of the virus' autochthonous transmission in North America, and a substantial decrease of its incidence in the rest of the Americas (1,2).European countries, including France, have set themselves the goal of eradicating measles by 2007 (3).In Finland, no autochthonous cases of measles have been recorded since 1996 (4).
Is France on the way to achieving this goal?To answer this question, this article presents an overview of the epidemiology of measles in France.

Evolution of the immunisation policy
The vaccine came on the market in 1966 in France and was introduced into the vaccination schedule in 1983, for children ➤ ➤ vaccination contre les oreillons et la rubéole y était associée.
➤ between the ages of 12-15 months.Three years later, mumps and rubella vaccination was added to the measles vaccine.
In 1996, a second dose, justified by the elimination prospect of the disease was introduced for children aged 11-13 years.This second dose enables us to protect children who have missed vaccination, and also those for whom it was not effective (there are failure rates of between 5 and 10% after the first dose).This prevents the accumulation of nonprotected subjects, thereby hampering the emergence of clusters.Modelling studies (5) have shown that the disease could be eliminated faster if the second dose were given to children at a younger age, and recommendations are now to give it to children between 3 and 6 years of age (6).
Up to the age of 6, two doses are administered.After 6 and until 13, only one dose is offered to those who have never been vaccinated.Vaccination is carried out mostly by doctors in private practice.Campaigns are carried out each year to promote the triple vaccine for measles, mumps and rubella (MMR).The focus of the 1999 campaign was that the vaccine was available free of charge.This action gave the vaccine a priority status by making it more accessible.

Vaccination coverage surveillance
Since 1985, data on vaccination coverage have been collected and managed by the Direction de la Recherche des Etudes de l'Evaluation et des Statistiques (DREES) at the Ministry of Health.
During the last survey conducted in 2000-01, coverage for six year olds was 90%.This shows that catch up vaccination was incomplete for children over the age of 2 years (8).

Surveillance of morbidity
In 1945, measles was made a notifiable disease, but this notifiable status was suspended in 1986 because of the low level of notifications being made by doctors.
The Sentinel network, created by the INSERM U444 has been monitoring measles since 1985 (9).It is composed of around 300 volunteer general practitioners, who report electronically the number of measles cases observed during their consultations each week.The case definition, comparable to the one used by the Centers for Disease Control and Prevention (CDC) (10) includes generalised rash of more than three days, and a fever above 38.5 °C, associated with coughing, coryza or conjunctivitis.
Since its creation, the network doctors have reported more than 8000 cases of measles.National incidence, extrapolated from this data, has decreased considerably, from nearly 300 000 cases per year in 1985 to around 10 000 cases in 2000 (figure 3).
At the same time, the Sentinel network observed an increase in the age of the cases.The proportion of patients over 10 years old rose from 13% in 1985 to 48% in 1997.The proportion of vaccinated cases also increased from 3% to 58% over the same period.This increase primarily reveals the improvement of vaccination coverage.Until 1998, these rates were calculated from more than 200 cases described by the network doctors.In 1998, this number was reduced to around fifty cases, which has made data interpretation very difficult.
Surveillance relies on clinical data alone, and presently no laboratory confirmation of cases or procedure for the typing of strains exist.Alongside this surveillance system, data on morbidity has been reported during the investigation of clusters or small outbreaks.From 1995 to 2000, seven investigations were led by or reported to the Institut de Veille Sanitaire.Vaccine efficacy, calculated in two primary schools and a high school on clinical and biologically confirmed cases has always been over 92% (11,12).

Surveillance of mortality
Data on mortality is collected by the epidemiological centre for the medical causes of deaths (Centre d'Epidémiologie sur les causes médicales de décès, CépiDc).This data has been available since 1979.
Only information regarding the main causes of death has been analysed.A detailed study including all the recorded causes of deaths (principal, associated and immediate) showed that only two deaths were not included in the 35 main causes registered from 1995 and 1998.➤
Mortality has decreased steadily, from around 30 deaths per year in the 1980s to less than 10 per year at the present time.The proportion of deaths in children under five years of age decreased from over 50% in 1979 to less than 5% in recent years (figure 4).
In the same period, the main causes of death also changed.From 1979 to 1987, causes other than encephalitic ones represented the majority of deaths, bronchopneumonias being responsible for 34% of the other causes.Since 1987, encephalitis is the first reported cause of deaths, whether it is late encephalitis occurring a few years after measles (subacute sclerosing panencephalitis) or acute immediate or delayed post-measles encephalitis occurring a few days or months after measles.

Serological survey
A European seroepidemiological survey (13) (ESEN) was carried out in seven countries.Immuno-globulins G of nearly 29 000 serum samples were analysed with standardised titration methods (14).In metropolitan France, serum samples came from subjects whose blood samples were drawn in medical laboratories.The choice of serum samples was made according to predefined quotas by age group, sex, and location to ensure a satisfactory representativeness without random sampling (15).
In France, about 3500 serum sampled mainly in 1998 were analysed (13) (figure 5).Results for measles showed an important rate of seronegative children (7% in those aged between 5 and 19 years).
Rates of seronegatives observed in all age groups under 20 are higher than seronegativity thresholds proposed by the WHO European region to reach the objective of measles elimination.Serological data have also confirmed that the half southern metropolitan France, as defined in figure 2 was less protected than the northern half.
Le vaccin utilisé en France est de bonne qualité puisque les mesures d'efficacité vaccinale faites lors d'investigations de cas groupés ont toujours été au-delà de 90 %.Par contre, la couverture vaccinale n'est pas assez élevée pour éviter l'accumulation de sujets susceptibles comme le confirment les études sérologiques et l'éclosion de foyers épidémiques.La France est encore dans la phase « lune de miel », nom donné quand la baisse des cas fait croire à une maladie en voie de disparition alors que se forment, sans aucune manifestation, des poches d'individus non-protégés, source des prochains foyers épidémiques.Les efforts de vaccination ont permis une forte baisse de l'incidence sans pour autant éviter la formation à bas bruit de ces poches.Les activités de rattrapage importantes de ces dernières années ont probablement réduit ces poches de susceptibles sans pour autant les faire disparaître.Les différences géographiques sont importantes et en parallèle des campagnes de promotion nationale, chaque département doit s'attacher à identifier les populations les moins vaccinées pour pouvoir y remédier, recherche qui est rendue possible par la mesure des couvertures vaccinales, l'identification des cas groupés et la notification de tous les cas.L'amélioration des couvertures doit s'accompagner d'un changement du système de surveillance.En effet, la baisse d'incidence rend les estimations du réseau Sentinelles peu précises.En 2001, l'incidence a été extrapolée ➤ This survey is carried out by a list of GPs and specialists, from which is selected a representative sample stratified according to the doctors' region and activity level.During a period of seven days, the doctor reports a set of information on all his or her patients, as well as copies of all prescriptions.The results are then extrapolated to all French practitioners, yielding national incidence estimates.
In children under the age of two and a half years, the rate of prescription of the triple vaccine has remained stable since 1996, data confirmed by the stagnation of vaccine coverage at 24 months.The rate of prescription has increased in children aged between two and a half and seven years of age since 1997, when the age for the second dose was lowered, yielding a coverage for the second dose of about 50%.Finally, in children over seven years of age, prescriptions have increased widely, although it has been impossible to dissociate vaccines administered as late second doses from those corresponding to first doses in non-vaccinated children.The decrease of cases registered by the Sentinel network in recent years favours the second hypothesis (figure 6).

Discussion
Since 1983, promotion of measles vaccination has been accompanied by a 97% reduction in morbidity and a 60% reduction in mortality.These encouraging results are not equivalent to elimination.Stable and suboptimal coverage around 84% causes a shift in the age of patient, and the continuation of disease transmission within the country.
Because of the reduced circulation of the virus, measles cases now occur in older patients.But complications are more frequent and case fatality ratio is higher in adults, with acute encephalitis being the major cause of death (17).This age shift in cases could therefore partly explain the slower decrease of mortality compared to morbidity, a stable number of deaths caused by post-measles encephalitis since 1988, and the increase in the age of the deceased patients.Deaths caused by subacute sclerosing panencephalitis occur about ten years after infection by the measles virus, and do not reflect the current incidence of the disease, although they illustrate the severe potential of a disease that is too often considered to be totally benign.
These paradoxical effects of vaccination linked to the age shift can be fought only by eliminating the disease, and this requires a vaccine coverage of more than 95% and the administration of two doses of an efficient vaccine.
The vaccine used in France is of good quality considering that the measures of vaccine efficacy carried out in investigations of clustered cases have always been over 90%.However, the coverage reached is not high enough to avoid the accumulation of susceptible subjects, as confirmed by serological surveys and the occurrence of epidemic clusters.France is still in the 'honeymoon' period, the name given when the decrease of cases suggests that a disease is disappearing, yet without any sign, groups of unprotected individuals are forming the source of future outbreaks.Vaccination efforts have allowed an important decrease of incidence without preventing the silent formation of these groups.Important catch up activities in recent years have probably reduced these groups of susceptible subjects without causing them to disappear.Geographical differences are important, and alongside national promotional campaigns, each department must strive to identify the populations with the lowest vaccination coverage in order to find solutions.This search is made possible by the measurement of local vaccine coverage, the identification of clusters, and notification of all cases.
Improving vaccine coverage must be accompanied by a change in the surveillance system.The decrease of incidence makes the estimates from the Sentinel network very imprecise.In 2001, incidence was extrapolated from 22 reported cases; the 95% confidence interval was 2870-14050 (18).The decrease in the number of cases also ➤ ➤ decreases the positive predictive value of the clinical definition.In England and Wales, 2466 cases of measles were reported from the clinical definition in 2000, 1751 (71%) had a saliva sample, of which 74 (4 %) only were positive (19).
In the long term, surveillance tools should be those used by the countries which are most advanced in the elimination process (20), namely: exhaustive notification; wide clinical definition for high sensitivity allowing detection of all suspect cases; biological confirmation to improve specificity and to take into account real cases only; typing of strains to trace their origin; measure of vaccine coverage for each dose; and evaluation of the rate of susceptible population by modelling or serological survey.
As clusters affect communities, school doctors should be involved in the surveillance of the disease.Protocols for procedures to be followed when one or several cases occur will be developed to encourage case notification and investigation.
Research projects are led by the Sentinel network, in collaboration with the Public Health Laboratory Service (PHLS) in London and are related to the evaluation of the saliva diagnosis by IgM detection and virological typing of strains.The results will contribute to setting up the biological support to eliminate the disease.
In conclusion, France has improved its vaccine coverage, but the rate reached still permits transmission of the disease, and shifts the age of the cases to an age where complications and case fatality ratio are higher.To eliminate the disease, vaccine coverage over 95% with two doses will have to be reached.
All the tools can gradually be implemented to reach this objective.The main obstacle, however, is to convince the policy decision makers, the health professionals, and the public of the need to eliminate the disease.■

Méthodes
Un questionnaire a été envoyé par courriel aux douze instituts participants (de 10 pays) au projet (voir liste des participants) et également à l'Institut de microbiologie et d'immunologie de Slovénie qui n'était pas un membre officiel du groupe.Les groupes ont été inclus dans le projet parce qu'ils étaient intéressés par la biologie moléculaire du VDN ou du VHA, l'épidémiologie et/ou la sécurité alimentaire.Les informations ont été recueillies selon : 1) les techniques diagnostiques utilisées en routine dans le labora-T he importance of food and water in the transmission of Norwalk- like viruses (NLVs) and Hepatitis A virus (HAV) is increasingly recognised.Outbreaks spread by these modes have the potential to involve large numbers of people, be widely geographically spread and, perhaps, introduce new variants to an area.Virus typing, using modern molecular methods, has shown how contaminated food from one country has resulted in outbreaks in another country from imported food.The risk of hepatitis A (HAV) outbreaks is substantial due to waning population immunity as a consequence of safe water and sanitation in Europe.Norwalk-like viruses (NLVs) cannot be cultured in the laboratory, thus, diagnosis is based on electron microscopy (EM) or, increasingly, on molecular techniques.HAV is diagnosed routinely by the detection of virus-specific antibodies.
There are numerous reports of foodborne outbreaks of NLV and HAV, but the true incidence of these diseases and the contribution of these outbreaks to the disease burden remain unclear.
The "Foodborne viruses in Europe" research project was recently funded by the European Union to obtain better information on the impact of foodborne viruses within the EU.Groups were included in the network based on their interest and track record in viral gastroenteritis.The participant countries have networked their virological and epidemiological surveillance in order to detect transnational outbreaks as well as elucidate transmission routes.In order to facilitate the design of a European database, we surveyed all the participant countries of the "Foodborne Viruses in Europe" project about their laboratory capabilities as well as the databases that record cases and outbreaks of viral gastro-enteritis.

Methods
A questionnaire was sent by email to twelve participant institutions (from ten countries) in the "Foodborne viruses in Europe" project (see participants list) as well as the Institute of Microbiology and Immunology (Slovenia), which was not officially a member of the group.Groups were included in the project based on their interest in NLV and HAV molecular biology, epidemiology and/or food safety.Information was collected on 1) the diagnostic techniques used in routine laboratories 2) the number and types of databases used that store information on A database was defined as: any organised set of electronic or paper-based information on individual human cases or summaries of outbreaks.A national database was considered one designed to collect information from all geographic regions of a country, but no stipulations were made concerning the overall or regional levels of reporting.

Results
A completed survey questionnaire was returned from all 10 countries.

Human diagnostics
All countries have the capability to test for HAV immunoglobulin (Ig) in human serum but, the number of laboratories performing tests in each country varies widely (table).Laboratories in Germany, Spain, France, The Netherlands and Finland test for HAV RNA by reversetranscription polymerase chain reaction (RT-PCR).

Surveillance
All countries except France and Sweden have a national database of HAV cases (though a pilot surveillance has begun in France) (figure A).Of those countries that do maintain HAV case databases, all contain information from laboratory reports except for Germany.England and Wales, Italy, Finland, and the Netherlands maintain additional databases that receive reports from primary care doctors and other health care institutions.In addition, HAV databases derived from a range of special studies also exist: an outbreak survey of public health physicians in 1998 (England and Wales), a pilot surveillance of HAV infection (France), and a survey on travellers before vaccination (Slovenia).

Environmental testing
Food can be tested for the presence of HAV genetic material by RT-PCR by laboratories in Finland, Italy, Spain, France, and Denmark.Laboratories in these five countries as well as England and Wales can test for HAV in water and drinks.

Human diagnostic capability
All countries use RT-PCR and EM to detect NLVs, though in some countries EM is now used as a last line of detection (Table ).Two laboratories in England and Wales use an enzyme immuno assay (EIA) based on recombinant NLV.

Surveillance
All countries, except Spain, maintain at least one (laboratory-based) database of cases of NLV.In the Netherlands databases also hold information from primary care and other healthcare providers.In addition, there are historical databases from surveys such as the Dutch NIVEL case-control study, and SENSOR cohort study as well as a study of outbreaks in the western regions of France.Also in France, historical databases exist from a study of children (1995-98).England and Wales, Germany, Finland, Slovenia and Denmark have national databases of NLV cases (figure B).
Databases of NLV outbreaks are kept in every country.These databases are all laboratory-based apart from one based on primary care data (France), and two others based on health care facility data in Denmark and The Netherlands.There are also databases from special studies of structured outbreak surveillance in England, and Wales and The Netherlands.Except for Italy, outbreak databases in every country are designed to collect national data (figure C), ➤ toire, 2) le nombre et le type de bases de données utilisées pour le stockage des informations sur les infections à VDN et VHA, et 3) les capacités d'investigation environnementale pour ces virus.

Analyses environnementales
En Finlande, en Italie, en Espagne, en France et au Danemark, les laboratoires peuvent détecter la présence des VHA dans la nourriture par RT-PCR.Les laboratoires de ces cinq pays, plus ceux d'Angleterre et du Pays de Galles, sont capables de faire cette analyse dans l'eau et les boissons.
Des bases de données sur les épidémies de VDN sont maintenues dans chaque pays.Elles sont toutes basées sur les laboratoires, à part celle ➤ ➤ although this does not necessarily mean that all outbreaks that are investigated are then reported.In France and Denmark, only food-or waterborne outbreaks were reported.

Environmental testing
Laboratories that can test human stools also have the ability to test food animal stools, although only Finland, England and Wales, France, Denmark, and The Netherlands perform such tests.Most countries also have laboratories that test foods (namely shellfish) as well as water and drinks (see table ).

Discussion
This inventory of laboratory capacity is the first step towards harmonisation of laboratory and surveillance of foodborne viruses in participating countries.Both the capability and the infrastructure of laboratories testing for foodborne viral pathogens in Europe vary widely.The sharing of samples and experience through the Foodborne Viruses in Europe consortium will however make these diverse capabilities complementary.Some countries have a more centralised diagnostic service with nearly all testing of foodborne viruses concentrated in one or two laboratories.In other countries, such as England and Wales, Germany, The Netherlands, and Italy, diagnostics are provided in many laboratories.
Although the majority of countries in this survey have laboratories that can test for HAV infection in humans, many do not test for HAV in food, water or drinks and therefore, food and waterborne infections cannot be microbiologically proven without international assistance.All countries can test human faeces for NLV by RT-PCR and EM.In comparison to EM, the RT-PCR is a substantially more sensitive diagnostic tool, able to detect virus up to two weeks after infection .Thus, the use of RT-PCR could increase the laboratory ascertainment of cases and outbreaks of NLV.Due to the genetic diversity of NLVs it has been difficult to develop a sufficiently sensitive and specific assay.In order to harmonise diagnostic methods used across the network, laboratories have tested a representative panel of stool samples using the different assays used among the network participants.The results of this evaluation will be presented elsewhere.
Testing of shellfish can be performed in most countries though satisfactory methods have not been standardised.There are no established tests for assaying other foods, and one of the aims of the "Foodborne Viruses in Europe" is to develop such techniques.Laboratories in Finland, The Netherlands, England and Wales, Denmark, and France have used their facilities to test food animals for NLV, thus capturing the possibility of zoonotic infection.Though transmission of NLV from animal to humans has not been demonstrated, evidence of the potential has been mounting in recent years.NLV genes have been detected in pigs and cattle, and the genetic material of bovine caliciviruses were shown to be very similar to human NLV.
All countries have at least one database of NLV outbreaks and all but Italy have a database with national coverage.Ascertainment of NLV gastroenteritis by routine surveillance has been shown to be poor since the condition is typically mild in that it does not cause the affected individuals to seek medical attention.Though the level of ascertainment may vary widely, we believe that an international database that captures outbreaks would be most appropriate for the "Foodborne Viruses in Europe" network.
Diagnostics and, therefore, surveillance of foodborne viruses are rapidly evolving fields.Molecular techniques used to detect viral RNA (by RT-PCR assay) and virus particles (EIA) are increasingly used in diagnostic laboratories.And, many of the national databases and surveillance networks that are referred to in these reports have been in existence for a short period of time.For example, reporting of NLV cases in Germany to a central database began as recently as January 2001.
Through the use of molecular typing, a transnational foodborne outbreak of NLV where contaminated raspberries from Slovenia caused infections in Europe and Canada has been described.Large scale food-and waterborne outbreaks have been documented a number of times but this proof of a transnational outbreak demonstrates the potential health impacts of the diffuse distribution of modern foodstuff industry.Through the creation of an international epidemiological and molecular database, the "Foodborne Viruses in Europe" project will study how common such outbreaks are and which are the transmission routes associated with most outbreaks.It will also enable the detection of epidemics and of epidemic variant viruses at an early stage.■ Les diagnostics et donc, la surveillance des virus d'origine alimentaire sont des domaines en pleine expansion.Les techniques moléculaires pour détecter l'ARN viral (par RT-PCR) et les particules virales (EIA) sont de plus en plus utilisées par les laboratoires de diagnostic.De nombreuses bases de données nationales et les réseaux de surveillance mentionnés dans ces rapports n'existent que depuis peu.Par exemple, la déclaration des cas de VDN à une base de données centrale en Allemagne n'a commencé qu'en janvier 2001.
Les six souches humaines de S. Dublin provenaient de trois provinces d'Autriche (trois du Tyrol, deux de Haute-Autriche, et une de Vienne).Les deux patients de Haute-Autriche avaient été contaminés à l'étranger (au Vietnam et en Inde).La patiente de Vienne est née au Nigeria, elle était séropositive pour le VIH et a déclaré n'avoir effectué aucun voyage hors de Vienne depuis trois ans qu'elle séjournait en Autriche.Les patients du non-human) can be traced epidemiologically to two districts in the Tyrol.This concentration of cases can be explained by a particularly traditional aspect of cattle farming in this area, the alpine pasture.There is an increased risk of cross infection due to the communal keeping of animals from various farms.Infected cattle are a source of infection for humans, and transmission usually occurs from eating beef and drinking cows' milk.Using pulsed field gel electrophoresis and automated ribotyping, three out of five isolates from human infections could be traced to characteristic Tyrolean Dublin clones.Bacteriological screening for faecal carriage before the transfer of cattle from risk-herds to the alpine pastures and before the return from risk-pastures to the farms would be a possible starting point to prevent cross-contamination of large mixed herds and contamination of pasture through latently infected cattle.Appropriate research is necessary.
W idespread distribution of Salmonella enterica in the environ- ment, its prevalence in the global food chain, and its virulence and adaptability have an enormous impact on medicine, public health, and the world economy (1).Infection with non-typhoidal S. enterica most often results in self limited acute gastroenteritis.Between 1% and 4% of immunocompetent individuals with salmonella gastroenteritis have positive blood cultures for S. enterica (2).
Most salmonellae are widely dispersed in nature, and can be found in the gastrointestinal tracts of domesticated and wild mammals, reptiles, birds, and insects.Some S. enterica serovars such as Typhi are highly adapted to humans and have no other known hosts.Salmonella enterica subsp.enterica serovar Dublin is primarily adapted to bovines and is a rare cause of human illness (3).S. Dublin infection in humans is severe relative to other forms of salmonella infection.Classically, it produces a syndrome of sustained bacteraemia with fever.In 2000, one human case of S. Dublin infection was documented in Austria.The 71 year old patient originated from a small village in the Tyrol and required hospitalisation for 14 days.Reflecting its increasing political interest in the prevention of zoonoses, the Federal Ministry of Health asked for an epidemiological review of S. Dublin infections in Austria over the past ten years.

Epidemiological review
The database of the national reference laboratory for salmonella (Graz, Austria) was used to find all initial S. Dublin strains isolated in Austria between 1990 and June 2001.A total of 77 salmonella isolates were of serovar Dublin.Six strains originated from humans (two blood cultures, four stool specimens) and 64 strains from cattle (23 faecal specimens, 36 organs from deceased animals, five meat samples from slaughtered animals culled due to illness).Four isolates were from chamois (organs), one isolate from a water sample and one from clay, both taken from water ponds epidemiologically involved in zoonotic outbreaks.One strain was isolated from pooled faecal material at a small farm during routine screening before the slaughter of a chicken flock (approximately 20 birds).
Figure 1 presents the S. Dublin isolates in chronological order stratified into "human", "cattle", and "other" isolates.Figure 2 gives the geographical distribution (and place of infection if different from place of diagnosis).
The six human S. Dublin strains originated from three Austrian provinces (three from the Tyrol, two from Upper Austria, and one from Vienna).Both patients from Upper Austria acquired S. Dublin abroad (Vietnam and India).The patient from Vienna was born in Nigeria; she was positive for antibodies to HIV and denied any travel outside Vienna during her three years in Austria.The Tyrolean cases all came from the two eastern districts of Kufstein and Kitzbuehel, and could be linked to the consumption of unpasteurised cows' milk (two) and veal from domestic butchering (one).
Tyrol venaient tous des deux districts à l'est de la région, Kufstein et Kitzbuehel.Leur contamination était sans doute liée à la consommation de lait de vache non pasteurisé, pour deux d'entre eux, et à celle de viande de veau tué à la ferme, pour le troisième.In Tyrol and Salzburg, S. Dublin was found to be an endemic zoonosis.The isolate from Upper Austria, from an animal slaughtered because of illness at the abattoir in Linz was also found to be from an endemic case.It could be traced back to a small farm in the district of Ried im Innkreis, which lost three animals to disease within a few weeks.All animals were bred and reared at this farm, which had acquired two cows from the Tyrolean district of Kitzbuehel 15 years before.
The Enter-net salmonella database was used to compare incidences of S. Dublin infections in humans in Austria and 15 other Western European countries.In 2000 there were only 181 cases in the 16 European countries analysed in the database with a European rate of 0.4864 cases per million population.The national rates range from 0-3.3333 cases depending on the country: the rate for Austria is 0.1250 cases per million population.Enter-net data do not allow uncritical comparison of rates between countries because of the large differences in sensitivity and ascertainment of cases within these countries.

Laboratory review
Twenty-one S. Dublin isolates of human and nonhuman origin were available for further laboratory based subtyping to determine their clonal and therefore their epidemiological relatedness.Subtyping was performed using automated ribotyping (RiboPrinter®) and EcoRI as restriction enzyme (table) and PFGE pattern analysis (figure 3 and table) using the restriction enzymes XbaI, BlnI and SpeI according to Liesegang et al. (4) The clonal analysis of the selected and representative isolates of S. Dublin reveals the "Austrian" human isolates to be of heterogeneous nature; only two of the isolates were found to be clonal identical and indistinguishable from the isolates from calves, chicken, and ➤

Discussion
Salmonella enterica sérotype Dublin, adaptée aux bovins, est considérée comme la cause la plus courante d'infection à salmonelles chez le bétail, avec S. Typhimurium (5-7).Le bétail infecté est une source de contamination pour les humains, généralement par la consommation de viande de boeuf et de lait de vache (8,9).Les infections humaines à S. Dublin sont rares en Autriche.En 2000, l'Autriche était dans la moyenne européenne, avec une incidence de 0,1 cas par million d'habitants.Il est à noter que trois des isolats humains sur cinq se rapportaient à des clones caractéristiques de S. Dublin au Tyrol.Khaschabi et al ont analysé les isolats de boeuf du Tyrol en utilisant la PFGE et ont émis l'hypothèse d'une source clonale, due à l'absence de dissémination en dehors de la région (6).Nos données montrent qu'à l'évidence, une large majorité des cas de salmonelles sérotype Dublin en Autriche peut être liée à ce clone du Tyrol.Ainsi, la concentration de cas dans les districts de Kufstein et Kitzbuehel, qui comptent 24 684 et 20 965 têtes de bétail (en 1988), peut s'expliquer par le mode d'élevage traditionnel particulier au Tyrol, les pâturages alpins.Les animaux de plusieurs fermes sont regroupés en vastes troupeaux et menés sur les pacages de ➤ soil (table).Moreover, they seemed to be identical to S. Dublin isolates, but clearly different from respective German isolates (data not shown, see also 5,6).S. Dublin strains with running number 18 and 20 in the table, a patient from Linz who acquired S. Dublin infection in Vietnam and a patient from Vienna, who was an HIV positive native Nigerian, respectively, have been identified to be of different clonal nature from the Tyrolean type and therefore of different epidemiological relationships.The human S. Dublin strain from 1993 (running number 21) originated from a different Tyrolean cattle clone, no longer prevalent among the isolates from 2000 (see 5,6).

Discussion
Salmonella enterica serovar Dublin, a bovine-adapted serovar, is considered to be the most common cause of salmonella infection in cattle along with S. Typhimurium (5-7).Infected cattle are a source of infection for people and transmission usually occurs from eating beef and drinking cows' milk (8,9).In Austria, human S. Dublin infections are rare.In 2000, Austria was within the European mean with an incidence of 0.1 cases per million inhabitants.Interestingly, three out of five available isolates from human infections could be traced to characteristic Tyrolean S. Dublin clones.Khaschabi et al examined Tyrolean beef isolates using PFGE and postulated a clonal source due to lack of spread outside the region (6).Our data show that the vast majority of all Austrian serovar Dublin salmonella cases can be traced to this Tyrolean clone.Hence the concentration of the cases in the districts of Kufstein and Kitzbuehel, with cattle populations of 24 684 and 20 965 animals (in 1988), can be explained by a particularly traditional aspect of cattle farming in the Tyrol, the alpine pasture.Cattle from several farms are herded together in large numbers to graze on high alpine meadows during the summer months.Thus there haute montagne pendant les mois d'été.Cette garde commune du bétail génère un risque accru d'infections croisées.L'infection enzootique à salmonelle du bétail adulte est, de ce fait, une maladie transmissible typique des alpages.S. Dublin peut survivre pendant plus d'un an dans l'environnement avec des conditions climatiques variées (10,11).Les signes cliniques se manifestent généralement par une diarrhée aiguë responsable de la déshydratation, avec un taux de mortalité élevé.Un avortement spontané, habituellement dans le septième ou huitième mois, peut être le seul symptôme de l'infection (12).Un cas d'infection à S. Dublin n'est pas facile à identifier rétrospectivement, car les excrétions de salmonelles sont généralement transitoires et les titres des anticorps sériques agglutinants tombent souvent à un niveau très faible peu après l'avortement (12).L'infection à S. Dublin chez le bétail se caractérise aussi par le nombre important de porteurs (13)(14)(15).Dans les districts de Kufstein et de Kitzbuehel, les vétérinaires connaissent bien le problème de l'infection à S. Dublin chez le bétail (5)(6)(7).Comme la loi autrichienne concernant les contagions animales, écrite en 1909, ne précise aucune réglementation sur les infections à salmonelles dans le bétail, les éleveurs doivent supporter les conséquences et les coûts liés à la maladie.En Autriche, la déclaration des cas animaux de salmonellose n'est même pas obligatoire ; ce n'est que dans le cadre de l'inspection alimentaire que les résultats positifs doivent être rapportés au vétérinaire responsable du district.Il serait tout à fait souhaitable d'instaurer une législation, comme la loi allemande sur les infections à salmonelles chez le bétail, pour favoriser l'instauration de mesures médicales préventives (16).
Plusieurs épidémies européennes des années précédentes, associées à des produits laitiers, montrent que les infections à S. Dublin chez le bétail représentent sans aucun doute un risque ➤ is increased risk of cross infection due to the communal keeping of animals from various farms.The enzootic salmonella infection of adult cattle is therefore a typical contagious disease in grazing pastures.S. Dublin can survive for more than one year in the environ-ment under various climatic conditions (10,11).Clinical manifestation is usually characterised by severe diarrhoea and resulting dehydration with a high rate of mortality.Spontaneous abortion, usually in the seventh or eighth month, can be the only symptom of infection (12).Retrospective identification of a case of S. Dublin may prove difficult as excretion of the organism is usually transient and the serum agglutinating antibodies frequently fall to low titres soon after the abortion (12).Another characteristic of S. Dublin infection of cattle is its high carrier rate (13)(14)(15).
The problem of S. Dublin infection in cattle is well known to veterinarians in the districts of Kufstein and Kitzbuehel (5-7).As Austrian law concerning animal contagions, written in 1909, does not specify any regulations on salmonella infection in cattle, the owners of the cattle are left to carry the consequences and the costs thereof.In Austria, cattle afflicted with salmonella infections do not even have to be reported to the authorities; only within the framework of meat inspection do positive results have to be reported to the veterinarian in charge of the district.Legal rulings, such as the existing German law concerning salmonella infection in cattle, would be most desirable to facilitate preventive medical measures (16).
For consumer protection it would be necessary to create regulations to compensate cattle farmers for their entire expenses and losses (spoiled and rejected animals) due to S. Dublin with the aim of achieving optimal co-operation between the owners of the animals and the official veterinarians in charge.At the very least, costs of laboratory examinations (faecal specimens examined for salmonella: 15 ATS or 1,1 € per specimen should be covered.For financial reasons, searches for sources of infection are generally being avoided at present.Bacteriological screening of herds with a known history of S. Dublin infection would be a start to prevent future contamination of alpine pastures through latently infected cattle excreting potentially infectious faeces.Bacteriological screening for faecal carriage before the return of cattle from pastures known to be connected with infections might be able to prevent cross-contamination of large mixed herds. Various European outbreaks connected to dairy products that occurred during the previous years demonstrate that S. Dublin in cattle certainly poses a preventable health risk for humans (8,9,17).The exact economic loss as a result of S. Dublin infections in humans and animals is not known.The official cost of the 14 days hospitalisation of the sole Austrian patient diagnosed with S. Dublin infection in the year 2000 came to a total of ➤

Figure 4 Nombre
Figure 4 Nombre et causes de décès annuels par rougeole, et proportion d'enfants de moins de 5 ans (moyenne sur 3 ans) 1979-1998, France / Number and causes of yearly deaths due to measles, and rate of children under 5 years of age (median over 3 years), 1979-1998, France

Figure
Figure Bases de données nationales pour les infections virales d'origine alimentaire / National database of foodborne viral infections

Figure 2 Figure 1
Figure 2 Répartition géographique des isolats de S. Dublin et lieu de la contamination (si différent du lieu de diagnostic), Autriche 1991-juin 2001 / Geographical distribution of S. Dublin isolates and place of infection (if different from place of diagnosis), Austria 1991-June 2001

This report describes a survey of national laboratory capa- bilities of diagnostics and surveillance databases for foodborne viruses among the "Foodborne Viruses in Europe" consortium. All the countries have laboratories that can test for HAV antibody in human serum. Eight of the ten surveyed Euro- pean countries maintain a national database of HAV cases. Food can be tested for the presence of HAV in Finland, Italy, Spain, France and Denmark. All surveyed countries have at least one laboratory that tests for Norwalk-like virus (NLV) by reverse transcriptase- polymerase chain reaction and all also have the capability to use electron microscopy. Five countries maintain a national database of NLV cases and nine maintain a national database of NLV outbreaks. Almost all participant countries have laboratories that can test for NLV in food items including shellfish. Tableau / Table Bases de données et capacités des laboratoires pour dépister les virus d'origine alimentaire au sein du Consortium en Europe / Databases and laboratory capability for foodborne viruses among the consortium on Foodborne Viruses in Europe Cas humains Surveillance Analyses environnementales Human cases Environmental testing Nombre de laboratoires analysant Nbre total de bases de données Nombre de laboratoires analysant Number of laboratories testing for recueillant des informations sur Number of laboratories testing for Total number of databases collecting information on
VHA / HAV= Virus de l'hépatite A / Hepatitis A virus VDN / NLV= Virus de Norwalk / Norwalk-like virus EIA = enzyme immuno assay ; ME/EM=microscopie électronique / electron microscopy; RT-PCR= reverse transcriptase polymerase chain reaction A une base de données nationale / Have national database Pas de base de donnés nationale / No national database B.

Bases de données nationales pour les cas de VDN / National NLV case databases C. Bases de données nationales pour les épidémies de VDN / National NLV outbreak databases A. Bases de données nationales pour les cas d'hépatite A / National hepatitis A case databases 62
e rôle prépondérant des aliments et de l'eau dans la transmission des virus de Norwalk (VDN) et dans celui de l'hépatite A (VHA) est de plus en plus reconnu.Les épidémies qui se répandent ainsi ont le potentiel de toucher un grand nombre de personnes, de se répandre géographiquement de manière importante, et peut-être même d'introduire de nouveaux variants dans une région.Le typage des virus par des méthodes Foodborne Viruses in Europe » des informations sur les capacités techniques de leurs laboratoires et sur les bases de données recensant les cas et les épidémies de gastro-entérites virales.