Vaccine effectiveness in preventing laboratory-confirmed influenza in primary care patients in a season of co-circulation of influenza A ( H 1 N 1 ) pdm 09 , B and drifted A ( H 3 N 2 ) , I-MOVE Multicentre Case – Control Study , Europe 2014 / 15

M Valenciano 1 2 , E Kissling 1 2 , A Reuss 3 , C Rizzo 4 , A Gherasim 5 , JK Horváth 6 , L Domegan 7 , D Pitigoi 8 , A Machado 9 , IA Paradowska-Stankiewicz 10 , A Bella 11 , A Larrauri 12 , A Ferenczi 13 , Joan O ́Donell 7 , M Lazar 14 , P Pechirra 15 , MR Korczyńska 10 , F Pozo 16 , A Moren 1 , on behalf of the I-MOVE multicentre case–control team 17 1. Epidemiology Department, EpiConcept, Paris, France 2. These authors contributed equally to this manuscript 3. Department for Infectious Disease Epidemiology Respiratory Infections Unit, Robert Koch Institute, Berlin, Germany 4. Istituto Superiore di Sanità, Rome, Italy 5. National Centre of Epidemiology, Institute of Health Carlos III, Madrid, Spain 6. Department of Public Health, Strategic Planning and Epidemiology, Office of the Chief Medical Officer, Budapest, Hungary 7. Health Service Executive-Health Protection Surveillance Centre, Dublin, Ireland 8. University of Medicine and Pharmacy ‘Carol Davila’, ‘Institutul National de Cercetare-Dezvoltare pentru Microbiologie si Immunologie ́Cantacuzino ,́ Bucharest, Romania 9. Department of Epidemiology, National Institute of Health Dr. Ricardo Jorge, Lisbon, Portugal 10. Polish National Institute of Public Health, Warsaw, Poland 11. Istituto Superiore di Sanità, Rome, Italy 12. National Centre of Epidemiology/ CIBER Epidemiología y Salud Pública (CIBERESP), Institute of Health Carlos III, Madrid, Spain 13. Department of Public Health, Strategic Planning and Epidemiology, Office of the Chief Medical Officer, Budapest, Hungary 14. ‘Institutul National de Cercetare-Dezvoltare pentru Microbiologie si Immunologie ́Cantacuzino ,́ Bucharest Romania 15. Department of Infectious Diseases, National Institute of Health Dr. Ricardo Jorge, Lisbon, Portugal 16. National Centre for Microbiology, National Influenza Centre Institute of Health Carlos III, Madrid, Spain 17. The members of the team are listed at the end of the article


Introduction
In February 2014 each year, the World Health Organization (WHO) provides recommendations for the composition of the northern hemisphere vaccines, based on information from the WHO Global Influenza Surveillance and Response System.In 2014, the WHO vaccine strain selection committee recommended that the 2014/15 northern hemisphere influenza vaccine should include the same components as in 2013/14: an A/California/7/2009 (H1N1)pdm09-like • With interval between onset of symptoms and swabbing >7 days (n=137) • Excluding patients presenting before ISO week of any influenza case and after ISO week of last influenza case after which there are two consecutive weeks of no cases (weeks of symptom onset, by country) (n=62) N=6,579 ; cases of any influenza: 3,437; controls: 3,142 a This includes 15 influenza B + A(H3N2) co-infections and eight influenza B + A(H1N1)pdm09 co-infections.Note that numbers of cases come from influenza type/subtype specific databases.Some cases are excluded due to their restriction criteria.Any influenza A non-typed cases are dropped from analysis.

Figure 3
Pooled crude and adjusted seasonal vaccine effectiveness against laboratory confirmed influenza by influenza type/subtype, and by season of virus, an A/Texas/50/2012 (H3N2)-like virus, and a B/ Massachusetts/2/2012-like virus [1].
In September 2014, the WHO reported the emergence of two new influenza virus genetic clades for A(H3N2), clade 3C.2a and 3C.3a [1].These clades had first circulated in Europe during the 2013/14 influenza season [2].
In December 2014, the United States (US) Centers for Disease Control and Prevention (CDC) issued a Health Alert reporting that 52% of the A(H3N2) viruses circulating were antigenically different from the A(H3N2) component of the northern hemisphere 2014/15 influenza vaccine.CDC recommended the use of antiviral medications where indicated for the treatment and prevention of influenza, as an adjunct to vaccination [3].Concordant with the reports of the drifted A(H3N2) viruses, in January 2015, the US, Canada and the United Kingdom (UK) reported low influenza vaccine effectiveness (VE) against A(H3N2) [4][5][6].Canadian results suggested that VE against influenza A(H3N2) among individuals who had been vaccinated in both 2013/14 and 2014/15 seasons was lower than among those who were only vaccinated in 2014/15 [5].
In Europe, the influenza season started later than in the US and Canada.Increased influenza activity in Europe was first reported in early January 2015, with a predominance of A(H3N2) but with influenza A(H1N1) pdm09 and B circulating as well [7].
For this seventh season of the Influenza Monitoring Vaccine Effectiveness in Europe (I-MOVE) multicentre case-control study we aimed to measure the 2014/15 effectiveness of the seasonal influenza vaccine against the three co-circulating viruses by age group and by vaccine type.In addition, due to the potential implications for vaccination policy we explored the effect of previous vaccinations on the current season VE.

Methods
Eight study sites (Germany, Hungary, Ireland, Italy, Poland, Portugal, Romania and Spain) participated in the test-negative 2014/15 multicentre case-control study.The methods have been described previously [7][8][9] and are based on the European Centre for Disease Prevention and Control (ECDC) generic case-control study protocol [10].Briefly, participating general practitioners (GPs) interviewed and collected naso-pharyngeal specimens from all (seven study sites) or a systematic sample (in Germany) of patients consulting for influenza-like illness (ILI) aged 60 (Germany, Poland, and three regions in Spain) or 65 years old (Hungary, Ireland, Italy, Portugal, Romania and three regions in Spain) and older and from a systematic sample of ILI patients in the other age groups.In Hungary, only patients aged 18 years or over were eligible for inclusion in the study.GPs collected clinical and epidemiological information as previously described [8].We included patients in the study who presented to the GPs more than 14 days after the start of the national vaccination campaigns and who met the European Union (EU) ILI case definition [11], were swabbed within seven days of symptom onset, and who had not received antivirals before swabbing.
Cases were ILI patients who were swabbed and tested positive for influenza virus using real-time reverse-transcription PCR (RT-PCR).Controls were ILI patients who tested negative for any influenza virus using RT-PCR.Cases and controls were not included in the influenza type/subtype-specific analyses if fewer than five type/ subtype-specific cases were reported by study site.
Influenza A cases of unknown subtype were excluded from the analysis.
For each study site and for each influenza type/subtype, the study period started on the week of onset of the first influenza case recruited and ended on the week of onset of the last influenza case after which there were at least two consecutive weeks with no further influenza positive cases.
We defined a patient as vaccinated if they had received minimum one dose of 2014/15 influenza vaccine at least 15 days before ILI symptom onset.We considered all other patients unvaccinated.GPs ascertained vaccination based on vaccination records or patient's self-report.
For each study site, we compared the odds of vaccination in cases and controls calculating the odd ratio (OR).We conducted a complete case analysis excluding patients with missing values for any of the variables in the model measuring adjusted VE.We carried out a one-stage model with study site as a fixed effect.We used Cochran's Q-test and the I 2 index to test the heterogeneity between study sites [12].
We used a logistic regression model to calculate VE including potential confounding factors: age (modelled as a restricted cubic spline with four knots or age group as a categorical variable depending on the analysis), sex, presence of at least one underlying chronic condition (including pregnancy and obesity where available) and date of symptom onset (modelled as a restricted cubic spline with four knots where sample size allowed).
To study the effect of 2013/14 vaccination on the 2014/15 VE, we conducted a stratified analysis using four categories: individuals unvaccinated in both seasons (reference category), vaccinated in 2013/14 only, vaccinated in 2014/15 only, and those vaccinated in both seasons.
We measured VE by age group (0-14, 15-59 and ≥60 years) and by type of vaccine (adjuvanted, egg-derived inactivated subunit, cell-derived inactivated subunit, egg-derived inactivated split virion).We excluded  study sites from the vaccine type analysis, where the given type of vaccine was not available.
We conducted four sensitivity analyses (i) restricting the study to patients swabbed less than 4 days after symptom onset, (ii) restricting to the population targeted for vaccination as defined in each country [23] (iii) excluding patients vaccinated < 15 days after symptom onset, (iv) calculating adjusted VE using a twostage model using random effects.
The respective country's National Influenza Reference Laboratories tested swab specimens for influenza by real-time RT-PCR assays.In Spain, other laboratories participating in the National Influenza Sentinel Surveillance System tested specimens.In each study site, a non-random selection of positive specimens or isolated viruses from positive specimens were subsequently sent to the corresponding National Influenza Centre, where influenza diagnosis was confirmed and viruses characterised either by sequencing the HA1 coding portion of the haemagglutinin gene (genetic characterisation) or by haemagglutination inhibition (antigenic characterisation).The criteria to select the specimens for genetic and antigenic characterisation varied by study site.
For the I-MOVE pooled analysis, the Spanish and Portuguese National Influenza Centres analysed the nt and amino acid sequences of the HA1 coding portion of the haemagglutinin gene and used the neighbourjoining method and the Kimura  and Kumar 2013).HA sequences from reference strains used in the phylogenetic analysis were obtained from the EpiFlu database of the Global Initiative on Sharing Avian Influenza Data (GISAID) (Table 1).

Results
Within the I-MOVE multicentre case-control study, the start of country-specific study periods ranged from week 41, 2014 (Germany) to week 3, 2015 (Poland), and the end from week 13, 2015 (Portugal) to week 19, 2015 (Germany).Study period duration ranged from 14 (Poland) to 31 (Germany) weeks.
Among the 7,992 ILI patients recruited, 6,579 ILI patients met the eligibility criteria including 3,142 testing negative for all influenza viruses.For the influenza type/subtype-specific analysis datasets, we included 1,828 influenza A(H3N2), 1,038 influenza B, 539 influenza A(H1N1)pdm09 (Figure 1).
The median onset date was 1 February for A(H1N1) pdm09, 1 February for A(H3N2), and 20 February for B cases (Figure 2).Forty-one percent of A(H3N2) cases were recruited in Germany, 44% of A(H1N1)pdm09 in Italy and 30% of B cases in Spain.
The median age was higher in influenza B cases (39 years) compared with influenza A(H3N2) and A(H1N1) cases (28 and 30 years respectively) and controls (31 years).
The proportion of patients swabbed more than three days after ILI onset was 15.9% among controls, and 10.3%, 13.5% and 15.9% among A(H3N2), A(H1N1) pdm09 and B cases respectively.
The proportion of patients belonging to the target group for vaccination, or with at least one chronic condition or with at least one hospitalisation in the previous 12 months was similar between influenza A(H3N2), A(H1N1)pdm09, B cases and controls.
Of the 735 vaccinated individuals, 620 (84%) had information on the vaccine type received; they were vaccinated with ten different brands.By vaccine type, 40% had received egg-derived inactivated subunit (used in all sites except in Hungary and Italy), 33% egg-derived inactivated split virion (used in all sites except in Ireland and Romania), 21% adjuvanted (used in Germany, Hungary, Italy and Spain) and 5% cellderived inactivated subunit vaccines (used in Germany and Spain).
After excluding patients with missing information (n = 833; 7%), we included 4,491, 2,920 and 3,730 patients in the complete case analysis of VE against influenza A(H3N2), A(H1N1)pdm09 and B respectively (Figure 1).
The I 2 was < 50% (p > 0.05) when assessing crude type/ subtype specific VE by study site and age group.Sample size among the 0-14 year-olds for the A(H1N1) pdm09 analysis was too small to carry out tests for heterogeneity.When assessing crude VE against A(H3N2) by study site among the target group for vaccination, the I 2 was 61.5% (p = 0.016).
The overall adjusted VE point estimate was similar to the adjusted VE among those swabbed less than 4 days of symptom onset (17.4%) and to the adjusted VE excluding individuals vaccinated less than 15 days after symptom onset (13.7%).The adjusted VE point estimate was higher when restricting the analysis to the target population (26.2%) (Table 2).The adjusted VE estimates using a two-stage random effects model were similar (within 6 % points) to the one-stage pooled analysis VE for all population and restricted to the target group for vaccination (Table 2).The twostage VE point estimate in the ≥60 year-olds was 10% higher than the one-stage VE but three study sites were excluded from the two-stage analysis due to their limited sample size.
Twelve viruses belonged to the group 3C.3a that harbours the T128A, R142G, A138S, N145S, F159S and N225D mutations.Nine of them had an extra mutation K276N at the antigenic site C. Fifty-eight viruses belonged to group 3C.2a and the only mutations identified were L3I, N144S, N145S, F159Y, K160T, N225D and Q311H -amino acid mutations that define the group.
By vaccine type, the adjusted VE point estimate was higher for the adjuvanted vaccine (79.8%) than for the egg-derived inactivated subunit and the inactivated split virion vaccines (53.0% and 51.5% respectively).We could not compute the VE for the cell-derived inactivated subunit due to small numbers (7 controls vaccinated and no cases vaccinated) (Table 4).
The overall adjusted VE point estimate did not vary when restricting the analysis to the target group for vaccination (53.6%), when excluding those vaccinated < 15 days (54.5%) before symptom onset and when using a two-stage pooled model (53.5%).It was 61.0% when restricted to those swabbed less than 4 days of symptom onset (Table 3).
There was less than 9% absolute difference between the overall adjusted VE point estimates and the VE in all sensitivity analyses (Table 3).The two-stage VE point estimate in the 0-14 years old was 15% lower than the one-stage VE point estimate but five study sites were excluded from the two-stage analysis due to their limited sample size.
Among 746 cases for which the lineage was available, 740 (99.2%) were Yamagata and six Victoria.

Discussion
The results of the I-MOVE multicentre case-control study suggest a low 2014/15 influenza VE against medically attended ILI due to A(H3N2) and a moderate VE against medically attended ILI due to A(H1N1)pdm09 or B.
The sample size of the I-MOVE multicentre case-control study for the 2014/15 season was one of the largest since 2008/09.We could estimate VE against the three circulating viruses.However, with the low influenza vaccination coverage in the participating sites, we still have limited statistical power for some subgroup analyses that provide important information for public health action like VE by previous vaccination or VE by type of vaccine.The current sample size is still too small to measure VE by vaccine product.
Measuring VE by study sites was not among the objectives of our multicentre study.In addition, as in previous seasons, study sites, sample size pending, are publishing their own results.However, even if not statistically significant, VE may differ between study sites.Differences in site-specific adjusted VE may be explained, among other factors, by variability due to the limited number of samples, unknown residual confounding, or different vaccines used.In future seasons we are confident that, with more resources, sample sizes should increase allowing for better adjustment and stratification including by vaccine brand.
Integrating virological and epidemiological information is essential to interpret VE estimates [5].For the  [8,9].They are lower than the final 2014/15 VE against A(H3N2) reported in the UK even if the proportion of drifted virus among those genetically characterised are higher in UK than in our study [14].VE against A(H3N2) was below 20% for all vaccine types with a lower point estimate for the cell-derived subunit vaccine.The effectiveness was lower in those vaccinated in both 2013/14 and 2014/15 than in those vaccinated only in the 2014/15 season.These observations are in line with the results of the 2014/15 early A(H3N2) VE estimates in Canada [5] and with those observed in previous studies [15][16][17].They are congruent with the hypothesis that prior immunisation may decrease the effectiveness of the vaccine and that this negative interference is more important when the antigenic distance is small between successive vaccine components but large between vaccine and circulating strain [18].These conditions were present in 2014/15 with an unchanged A(H3N2) vaccine component compared with the 2013/14 vaccine and with a mismatch between the vaccine and a high proportion of circulating strains.However, those results may be due to chance, or to bias.We need a much larger sample size to have higher precision in the estimates and to study the effect of prior vaccinations by age group.In our study, individuals vaccinated in both seasons are older than those vaccinated only in one season (median age 63 years and 50 years respectively).Unmeasured differences between individuals vaccinated in two consecutive seasons and those vaccinated only in one season may have affected the results.Previous vaccination was documented through GP records or patient self-reports and may be subject to error.Since neither the ILI patient nor the GPs knew if the patient was an influenza case we are confident that differential recall did not bias the results.If the results were not due to bias or to chance, concurrent immunological studies will be essential to better understand the biological mechanism behind, and the role of natural vs vaccineacquired immunity.
The VE estimates against influenza A(H1N1)pdm09 are similar to our results in previous seasons [7][8][9].The laboratory results indicate that the strains isolated from study participants were similar to the A(H1N1) pdm09 component of the 2014/15 influenza vaccine.As in 2013/14, we observed a lower VE among the elderly and higher among those aged 0-14 years old, however sample sizes were small in the age group analyses.The VE point estimates of the adjuvanted vaccines were higher but the small sample size in the analysis does not allow a comparison of effectiveness between vaccine types.
The VE against influenza B ranged from 41% to 62% in the overall population and was 56% in the target group for vaccination.Our estimates are similar to those reported by the UK [14].Nearly all viruses (99%) for which lineage was available were B/Yamagata and 98% of those characterised belonged to clade 3 that is antigenically similar to the vaccine virus.VE was similar by vaccine type with lower point estimate for cellderived inactivated subunit vaccines but the sample size is too low to interpret this observed difference.
The results suggested no effect of the 2013/14 vaccine and a slightly lower VE among those vaccinated in both seasons.
This is the third season we provide VE by vaccine type.A high proportion of vaccinated study participants (84%) had vaccine product documented.Even with one of the largest sample size since 2008/09, the numbers are still too low to measure adjusted VE by vaccine type and age group.The European Medicines Agency (EMA) requests that vaccine producers provide product-specific vaccine effectiveness [19].Taking into account the high number of vaccine products and the low vaccination coverage in countries participating in the study [20] the sample size to measure VE by vaccine product with high precision has to be much larger and substantial additional resources are needed.In a survey among I-MOVE partners to assess the feasibility of conducting product-specific VE in Europe (data not shown) most experts considered that in terms of resources allocation, providing precise estimates early in the season, by age group, by previous vaccination were of higher priority than measuring VE by product.
In summary, the 2014/15 results suggest a moderate effectiveness against influenza A(H1N1)pdm09 and B. The low effectiveness of the influenza vaccines against A(H3N2) observed again this season underlines the need to improve the A(H3N2) component of the vaccine especially among the target group for vaccination.This would be even more important if the observed negative effect of previous vaccination was confirmed.Since A(H3N2) virus is generally associated with more severe disease in the elderly and high-risk groups [21,22] and the vaccine is less effective against this influenza subtype, in seasons of A(H3N2) circulation early antiviral treatment should be recommended in these groups [3,6].
The effect of previous vaccinations is one of the questions that I-MOVE and other influenza VE teams in the US, Canada and Australia started to raise some years ago [17,[24][25][26][27].This is an important issue that may impact vaccination policy in Europe.They need to be addressed through international collaboration, a multidisciplinary approach and with long-term scientific independent studies.The I-MOVE multicentre casecontrol study should continue to increase the sample size and to strengthen the virological component of the study to contribute to answer these questions.

I-MOVE multicentre case-control team
Authors included in the I-MOVE multicentre case-control team (in addition to the 18 listed before and in alphabetical order of countries) • Germany: Valeria Alfonsi, Istituto Superiore di Sanità, Rome.
• Portugal: Ana Rodrigues, Department of Epidemiology, National Institute of Health Dr. Ricardo Jorge, Lisbon.
Number of influenza-like illness reports by case status and week of symptom onset, all influenza, target groups for vaccination, I-MOVE multicentre case-control study, Europe, influenza season 2014/15 (week 41/2014-week 19/2015) (n=6,524 a ) ILI: influenza-like illness; I-MOVE: Influenza Monitoring Vaccine Effectiveness in Europe, ISO: International Organization for Standardization.
• Dropping influenza-positive records of different type/subtype • Excluding patients presenting before ISO week of first type/subtype-specific influenza case and after ISO week of last type/subtype-specific influenza case after which there are two consecutive weeks of no cases (weeks of symptom onset, by country)

Table 1 a
Details of influenza haemagglutinin sequences obtained from GISAID used in the phylogenetic analysis, I-MOVE multicentre case-control study, Europe, influenza season 2014/15 (week 41/2014-week 19/2015) GISAID: Global Initiative on Sharing Avian Influenza Data.

Table 2
Details for influenza, A(H3N2), A(H1N1)pdm09 and influenza B cases and controls, I-MOVE multicentre case-control study, Europe, influenza season 2014/15 (week 41/2014-week 19/2015) (n=6,524 a ) Based on the complete case analysis: records with missing age, sex, chronic condition, vaccination status are dropped.Totals may differ between one-stage and two-stage models, as adjustment at study site-level may vary to the one-stage pooled model adjustment, resulting in different missing data dropped depending on included covariates.In addition different numbers of study sites may be included in each analysis due to sample size issues.Data adjusted for age (restricted cubic spline), onset date (restricted cubic spline), sex, chronic condition and study site.Exceptions are A(H3N2) all ages, where age groups(0-4, 5-14, 15-59 and ≥60 years)are used instead of restricted cubic splines.e Study sites include DE, ES, IT.HU not included in the 0-14 year old analysis, as no patients included aged <18 years.Sample size too low for IE, PT and RO.Study sites include DE, ES, HU, IE, IT, PT, RO.Sample size too low for PL.Crude VE for RO used in adjusted estimate, due to low sample size.Study sites include DE, ES, HU, IT, RO.IE, PL and PT not included due to low sample size.Crude VE for RO used in adjusted estimate, due to low sample size.Study sites include DE, ES, IE, IT, PL, PT, RO.HU not included in the 0-14 year old analysis, as no patients included aged <18 years.Data adjusted for age (restricted cubic spline), onset date (restricted cubic spline), sex, chronic condition and study site.Exceptions the A(H1N1)pdm09 analysis among the elderly, where data are adjusted for age (restricted cubic spline), onset date (restricted cubic spline), and study site only.j Study sites include DE, HU, IE, IT, RO, PL.ES and IE dropped from analysis due to small sample size.Study sites include DE, IT.ES, IE, PL, RO not included as sample size too low.HU not included in the 0-14 year old analysis, as no patients included aged <18 years.Study sites include DE, IT, RO.ES, HU, IE and PL not included as sample size too small.Crude VE for RO used in adjusted estimate, due to low sample size.Study sites include DE, IT.ES, HU, IE, PL and RO not included as sample size too small.Only crude VE available, due to low sample size.Study sites include DE, IT, RO.ES, HU, IE and PL not included as sample size too small.Crude VE for RO used in adjusted estimate, due to low sample size.Study sites include DE, ES, IT.IE, PL, PT and RO not included as sample size too low.HU not included in the 0-14 year old analysis, as no patients included aged < 18 years.
Of the 114 characterised A(H3N2) viruses, 107 (94%) were sequenced.Compared with A/Texas/50/2012, 17 viruses had the T128A, R142G and N145S mutations that define the group 3.C represented by A/ Samara/73/2013.Eight viruses had in addition the CI: confidence interval; DE: Germany; ES: Spain; HU: Hungary; IE: Ireland; I-MOVE: Influenza Monitoring Vaccine Effectiveness in Europe; IT: Italy; PL: Poland; PT: Portugal; RO: Romania; VE: vaccine effectiveness.a b c Crude VE adjusted by study site.d f g h i k l m n o Data adjusted for age (restricted cubic spline), onset date (restricted cubic spline), sex, chronic condition and study site.Exceptions the B analysis among the elderly, where data are adjusted for age (restricted cubic spline), onset date (restricted cubic spline), and study site only.p q Study sites include DE, ES, HU, IE, IT, PL, PT, RO.Crude VE for DE, HU, IE, PL and RO due to low sample size.r Study sites include DE, ES, HU, IE, IT, PL, PT, RO.Crude VE for HU, IE and RO due to low sample size.

Table 4
Pooled crude and adjusted seasonal vaccine effectiveness against laboratory-confirmed influenza by influenza type/ subtype, by vaccine type and by influenza vaccination status in 2013/14, I-MOVE multicentre case-control study, Europe, influenza season 2014/15 (week 41/2014-week 19/2015)Based on the complete case analysis: records with missing age, sex, chronic condition, vaccination status are dropped). a

Table 5
Influenza A(H3N2), A(H1N1)pdm09, B Yamagata, B Victoria viruses characterised by clade and study site, I-MOVE multicentre case-control study, Europe, influenza season 2014/15 (week 41/2014-week 19/2015) (n=291) Silke Buda, Department for Infectious Disease Epidemiology Respiratory Infections Unit Robert Koch Institute, Berlin.Beatrix Oroszi, Department of Public Health, Strategic Planning and Epidemiology, Office of the Chief Medical Officer, Budapest.Éva Herczegh, Influenza Virus Laboratory, National Center for Epidemiology, Budapest.• Ireland: Coralie Giese, EPIET, European Centre for Disease Control and Prevention, Stockholm; HSE-Health Protection Surveillance Centre, Dublin