With regard to the surveillance of infectious diseases such as influenza, the role of the European Union (EU) has become more important in recent years (6). The surveillance of influenza is a key element of the European influenza pandemic preparedness plan. An important task of surveillance is the early detection of influenza and the characterisation of potential pandemic strains from clinical specimens (7). To improve influenza surveillance in Europe, the EU has supported the creation of a Community Network of Reference Laboratories for Human Influenza (7) to accomplish several tasks, including the co-ordination of methods employed by the Member States for the diagnosis of influenza.

The European Scientific Working group on Influenza conducted an inventory in 1996 on the laboratory diagnostic and surveillance methods in 24 European countries (8). This study showed that the techniques used in influenza surveillance were heterogeneous and the performance of virological surveillance was therefore difficult to compare between countries. The methods used for the virological surveillance of influenza may have changed since 1996 and EISS wanted to have an update of the methods currently used for the testing of sentinel respiratory specimens in Europe. In addition, EISS wanted to know whether tests were routinely performed for the detection of other respiratory pathogens besides influenza. The inventory aimed to determine the status of virological methods routinely used by sentinel influenza surveillance networks participating in EISS during the 2001-2002 season.

Material and Methods
A questionnaire on virological methods used for influenza diagnosis and surveillance was developed and sent electronically to all EISS collaborating surveillance networks (Table 1) in June 2002. People that were responsible for collecting virological data in each network were asked to complete the questionnaire. If a network had more than one reference laboratory, respondents were asked to complete a single questionnaire. Twenty-one networks participated in the study.

The following topics were included in the questionnaire: specimen collection, laboratory diagnosis of influenza and tests for other respiratory infections besides influenza. The questions in the survey concerned data collected during the 2001-2002 influenza season. All 21 networks completed the questionnaire. Results based on sentinel data are presented for all networks except for Poland and Sweden. The results from Poland and Sweden are based on data from non-sentinel sources.

Results
Sentinel specimen collection and transport
Information on specimen collection is presented in Table 1. Most networks (12/20) collect nasal as well as throat swabs. The remaining networks collect either nasopharyngeal, or nasal, or throat swabs. In addition, three networks collect blood samples and one network nasal aspirates. Transport of the swabs occurred by mail in 16 networks and by courier in seven networks. Some networks used special delivery (Northern Ireland) or ambulance (the Czech Republic) for the transport of the swabs. The temperature at transport was ambient in 13 networks and 4ºC in five networks. The viral transport medium meant to preserve virus viability used was diverse, but usually contained antibiotics to inhibit growth of other microorganisms. Scotland used a lysis buffer specifically developed for preservation of nucleic acid, and therefore only suitable for PCR. The time delay in transport of the material from the sentinel physician to the laboratory varied between 0-120 hours for all networks; most networks reported a delay of 24-48 hours.

Methods used for sentinel virological surveillance
The methods routinely used by the EISS networks to isolate or identify the influenza viruses in sentinelrespiratory specimens are presented in Table 2.

All but two networks (the Netherlands and Scotland) used culture on MDCK cells for the detection of influenza viruses. Seven networks used culture on embryonated chicken eggs, and five networks used other cell lines in addition to MDCK cells. Diverse rapid techniques for virus detection are used, with RT-PCR most often used in the western countries and ELISA in the eastern countries.

The delay between specimen collection and the test result for typing (determination of influenza A or B) and subtyping (determination of H subtype and occasionally the N subtype) is shown in Table 2. The delay was variable and differed between EISS networks. A comparison of the delay in typing and subtyping is difficult to make since a variety of methods were applied to determine the type and subtype. For example, by using subtype specific PCR assays typing and subtyping can be done directly on the clinical specimen, whereas when typing and subtyping a virus isolate, the time needed to grow the virus is the defining factor.

For typing of influenza viruses the following methods were applied: PCR (11 networks), HAI (9 networks), IF (8 networks) and ELISA (7 networks). For subtypingsub typing the HAI assay was used in 15 networks. However, PCR was also used for subtyping in twelve networks. A total of nine networks applied more than one test to subtype influenza viruses. None of the five networks in eastern Europe used PCR, while 12 out of 14 networks that perform subtyping in western Europe used PCR (Table 2). Of these, eight networks used both HAI and PCR.

Testing sentinel specimens for other respiratory infections
Thirteen out of nineteen networks (the Czech Republic, England, France, Germany, the Netherlands, Northern Ireland, Portugal, Romania, Scotland, Slovenia, Spain, Switzerland, Wales) reported that they collect information on respiratory pathogens other than influenza virus in sentinel respiratory specimens. All thirteen networks collected information on respiratory syncytial virus (RSV), six networks collected data on adenovirus, five networks collected data on parainfluenzavirus and three networks collected data on rhinovirus. Three networks (England, the Netherlands and Slovenia) had information on other respiratory pathogens (e.g. coronavirus, Chlamydia pneumoniae, human metapneumovirus) (data not shown). Eleven networks reported that the sentinel swabs were tested for both influenza virus and RSV.

Discussion
The results highlight similarities in the specimen collection and transport procedures in the EISS networks. In most networks nose swabs as well as throat swabs were obtained and transported by mail to the laboratory. The laboratory methods used were heterogeneous, which confirms earlier findings (8). For virus culture, most networks used the same type of cells (MDCK), but for typing and subtyping of influenza viruses different methods (ELISA, HAI, PCR) were used. ELISA was more often used for typing and subtyping in eastern Europe and PCR was more frequently used in western Europe. Another important finding is that the majority of networks in EISS reported that they test sentinel swabs for other viruses (in particular RSV).

The type of respiratory specimen, the delay in the transport of swabs, the transport medium and the transport temperature are important factors that could potentially lead to an underestimation of the number of laboratory confirmed clinical cases of influenza reported by sentinel physicians. Our study has shown that most EISS networks used nose and/or throat swabs. In general, these are considered to be the right specimens for techniques such as culture and immunofluorescence (9). The transport of samples is advised at 4ºC or frozen at -70ºC (9). The outcome of our survey is that the specimens were often sent by post, at an ambient temperature and usually took 24-48 hours to reach the laboratory. This can be considered suboptimal, especially for virus culture. However, a study carried out in England and Wales found that clinical specimens sent by post provided good results when using multiplex RT-PCR techniques, although it is likely that there is some degradation of viral nucleic acid when specimens are transported this way (10). Another factor, the viral transport medium, should ideally include a balanced salt solution at neutral pH with protein stabilizers such as gelatine or bovine serum albumin (BSA) and antibiotics (9). The EISS networks used diverse media for the transportation of specimens, but in general these media met the mentioned demands.

All but one network used virus isolation on cell culture as the primary method for the detection of influenza virus. This approach is commonly used as the EISS laboratories characterise their virus isolates and/or send material to the WHO Collaborating Centre at Mill Hill for strain characterisation, an activity that is very important to map the spread of influenza globally and to establish the influenza vaccines in the southern and northern hemispheres each season. The reasons for using additional techniques, like PCR and ELISA, for detection were confirmation of the results, increased sensitivity and the detection of other respiratory pathogens such as adenovirus (e.g. in Slovenia, Spain and Switzerland).

The harmonisation of virological testing methods is an important objective of EISS. To initiate these efforts, a first Quality Control Assessment (QCA) was performed during the 2000-2001 season (11). Differences in virological results can be associated with the use of different laboratory techniques (e.g. PCR vs. cell culture (10, 12,13)) or differences in the application of the same laboratory technique (e.g. PCR). The first QCA, carried out in 16 EISS laboratories, found that the sensitivity of the RT-PCR in Europe varied widely (40-100% for influenza, 71-86% for RSV), depending on the laboratory (11). A second QCA was carried out during the 2002-2003 season and considerable improvements in the sensitivity rates were found (data not shown). The results of the first two QCAs, and QCAs planned in the future, will be used to further harmonise virological testing methods in EISS.

The finding that sentinel specimens were being tested for other respiratory infections is important for EISS, as many agents are associated with clinical symptoms of influenza-like illness and acute respiratory infection. An important pathogen that contributes to this burden of disease is RSV; in terms of mortality the role of RSV is suggested to be even greater than influenza B and influenza A/H1N1 (2). Our inventory found a large proportion of the networks testing sentinel specimens for RSV and EISS could therefore collect more detailed information on RSV activity in Europe. These findings have led to the creation of an RSV Task Group to explore how the surveillance of RSV could be better developed and further integrated into EISS.

In conclusion, sample collection and shipment are more or less similar whereas detection and (sub)typing methods are heterogeneous among the EISS networks. Despite this heterogeneity, results for detection and (sub)typing can be considerably improved when carefully controlled by external quality control, as the results of the two QCA studies showed. Further improvements may be made by a better harmonization and standardization of the applied methods. EISS will therefore take a number of actions within the framework of the recently created Community Network of Reference Laboratories for Human Influenza; these include the definition of basic tasks to be carried out by the laboratories, the preparation of standardised laboratory protocols and further QCAs.

This article was written on behalf of all EISS members: Alexandrescu V (RO), Aymard M (FR), Bartelds AIM (NL), Buchholz U (DE), Burguiere A-M (FR), Brydak L (PL), Cohen JM (FR), Domegan L (IE), Dooley S (IE), Falcao I (PT), Fleming DM (UK), Grauballe P (DK), Haas, W (DE), Hagmann R (CH), Havlickova M (CZ), Heckler R (DE), Heijnen M-L (NL), Hungnes O (NO), Iversen B (NO), de Jong JC (NL), Kennedy H (UK), Kristufkova Z (SK), Libotte M-L (BE), Lina B (FR), Linde A (SE), Lupulescu E (RO), Machala M (PL), Manuguerra J.-C. (FR), de Mateo S (ES), Meerhoff T (NL), Mosnier A (FR), Nolan D (IE), O'Neill H (UK), O'Flanagan D (IE), Paget WJ (NL), Penttinen P (SE), Perez-Brena P (ES), Pierquin F (BE), Pregliasco F (IT), Prosenc K (SI), Rebello de Andrade H (PT), Rokaite D (LT), Samuelsson S (DK), Schweiger B (DE), Socan M (SI), Thomas D (UK), Thomas Y (CH), Tumova B (CZ), Uphoff H (DE), Valette M (FR), Vega T (ES), van der Velden K (NL), van der Werf S (FR), Watson J (UK), Wilbrink B (NL), Yane F (BE) and Zambon M (UK).

Acknowledgements:
We would like to thank all national reference laboratories that participated in the study. In particular, we would like to thank Blaskovicova H (SK), Coughlan S (IE), Rimmelzwaan G (NL), Smith A (UK) Westmoreland D (UK) and Wunderli W (CH) for completing the questionnaire.