PDF hosted at the Radboud Repository of the Radboud University Nijmegen

Transcription

1 PDF hosted t the Rdboud Repository of the Rdboud University Nijmegen The following full text is publisher's version. For dditionl informtion bout this publiction click this link. Plese be dvised tht this informtion ws generted on nd my be subject to chnge.

2 Reserch rticles 2012/13 influenz vccine effectiveness ginst hospitlised influenz A(H1N1)pdm09, A(H3N2) nd B: estimtes from Europen network of hospitls M Rondy (m.rondy@epiconcept.fr) 1, O Luny 2,3,4, J Puig-Brberà 5, G Gefenite 6,7, J Cstill 8,9, K de Getno Donti 10, F Gltier 2,11,12, E Hk 6,7, M Guevr 8,9, S Costnzo 13, Europen hospitl IVE network 14, A Moren 1 1. EpiConcept, Pris, Frnce 2. French Clinicl Vccinology Network (REIVAC) 3. Cochin hospitl, Pris, Frnce 4. Institut ntionl de l snté et de l recherche médicle (Inserm), CIC BT 505 Cochin Psteur, Pris, Frnce 5. Vccines Reserch, FISABIO-Public Helth, Vlenci, Spin 6. Deprtment of Phrmcy, Unit of Phrmco-Epidemiology & Phrmco-Economics (PE2), University of Groningen, Groningen, the Netherlnds 7. Deprtment of Epidemiology, University Medicl Centre Groningen, Groningen, the Netherlnds 8. Instituto de Slud Públic de Nvrr, Pmplon, Spin 9. Centro de Investigción Biomédic de Epidemiologí y Slud Públic (CIBERESP), Mdrid, Spin 10. Deprtment of Infectious Disese, Ctholic University, Rome, Itly 11. Centre hospitlier régionl universitire (CHRU) Montpellier, Hôpitl Sint Eloi, Montpellier, Frnce 12. Inserm, CIC 1001, Montpellier, Frnce 13. Deprtment of Epidemiology nd Prevention, IRCCS Istituto Neurologico Mediterrneo Neuromed, Pozzilli, Itly 14. Members re listed t the end of the rticle Cittion style for this rticle: Rondy M, Luny O, Puig-Brberà J, Gefenite G, Cstill J, de Getno Donti K, Gltier F, Hk E, Guevr M, Costnzo S, Europen hospitl IVE network, Moren A. 2012/13 influenz vccine effectiveness ginst hospitlised influenz A(H1N1)pdm09, A(H3N2) nd B: estimtes from Europen network of hospitls. Euro Surveill. 2015;20(2):pii= Avilble online: Article submitted on 07 My 2014 / published on 15 Jnury 2015 While influenz vccines im to decrese the incidence of severe influenz mong high-risk groups, evidence of influenz vccine effectiveness (IVE) mong the influenz vccine trget popultion is sprse. We conducted multicentre test-negtive cse control study to estimte IVE ginst hospitlised lbortoryconfirmed influenz in the trget popultion in 18 hospitls in Frnce, Itly, Lithuni nd the Nvrre nd Vlenci regions in Spin. All hospitlised ptients ged 18 yers, belonging to the trget popultion presenting with influenz-like illness symptom onset within seven dys were swbbed. Ptients positive by reverse trnscription polymerse chin rection for influenz virus were cses nd those negtive were controls. Using logistic regression, we clculted IVE for ech influenz virus subtype nd djusted it for month of symptom onset, study site, ge nd chronic conditions. Of the 1,972 ptients included, 116 were positive for influenz A(H1N1)pdm09, 58 for A(H3N2) nd 232 for influenz B. Adjusted IVE ws 21.3% (95% confidence intervl (CI): to 50.6; n=1,628), 61.8% (95% CI: 26.8 to 80.0; n=557) nd 43.1% (95% CI: 21.2 to 58.9; n=1,526) ginst influenz A(H1N1) pdm09, A(H3N2) nd B respectively. Our results suggest tht the 2012/13 IVE ws moderte ginst influenz A(H3N2) nd B nd low ginst influenz A(H1N1) pdm09. Bckground Antigenic drifts of influenz viruses expose the popultion to new but relted influenz vrints on regulr bsis [1]. On the bsis of yerly revised composition of sesonl influenz vccines, the World Helth Orgniztion (WHO) considers nnul Influenz vccintion s the most efficient mesure ginst influenz [2]. Every yer, the sesonl influenz vccine licensure is obtined bsed on immunogenicity dt [3]. While these immunogenicity dt re thought to be vlid for helthy dults [4], the development of correltes of protection suited to vulnerble popultions is still to be chieved [5]. The popultion trgeted for influenz vccintion in Europe includes those t incresed risk of exposure to influenz virus s well s of developing severe disese, especilly disese resulting in hospitlistion or deth [6]. Trget groups for vccintion usully include dults over 59 or 64 yers of ge nd people of ny ge with certin underlying medicl conditions [7,8]. Mesuring influenz vccine effectiveness (IVE) in ech influenz seson is importnt for the following resons: to identify vccines types nd brnds with low IVE; to decide on lterntive preventive strtegies if erly estimtes of IVE re low (e.g. preventive use of ntivirls mong vulnerble individuls); nd to help decide on the next seson s vccine content. Repeted evidence of suboptiml IVE mong the popultion trgeted for nnul 1

3 Tble 1 Generic protocol dpttions in ech study site, hospitl-bsed influenz vccine effectiveness study, four Europen countries, 2012/13 Spin Protocol dpttion Frnce Itly Lithuni Nvrre Vlenci Additonl stff for the study Yes Yes No No Yes Services Emergency wrd Emergency wrd internl medicine unit Emergency wrd Infectious disese hospitl All Emergency wrd Vccine sttus scertinment Ptient Ptient or GP Ptient or GP Register Register nd orl Ascertinment of type of vccine used Ecologicl dt Individul dt Ecologicl dt Individul dt Individul dt Exclusion bsed on plce of residence No No No Yes Yes Inclusion of ptients unble to sign the consent form Yes Yes No Yes Yes Type of respirtory specimen Nsl Nsl nd phryngel One phryngel nd two nsl Nsl nd phryngel Nsl nd phryngel Dt entry vlidtion Study periods Influenz A(H1N1)pdm09 Influenz A(H3N2) Influenz B Coordintion tem Coordintion tem Coordintion tem Coordintion tem Double entry for lbortory results Weekly qulity checks Week 1, 2013 Week 2, 2013 Week 52, 2012 Week 7, 2013 Week 47, 2012 Week 10, 2013 Week 8, 2013 Week 9, 2013 Week 11, 2013 Week 15, 2013 Week 52, 2012 Week 3, 2013 Week 3, 2013 Week 4, 2013 Week 9, 2013 Week 14, 2013 Week 6, 2013 Week 13, 2013 Week 13, 2013 Week 12, 2013 Week 50, 2012 Week 5, 2013 Week 4, 2013 Week 50, 2012 Week 51, 2012 Week 13, 2013 Week 9, 2013 Week 15, 2013 Week 11, 2013 Week 15, 2013 GP: generl prctitioner. The Interntionl Orgniztion for Stndrdiztion s week numbers were used, to ensure consistency cross study sites. influenz vccintion would lso further dvocte the need for vccines tht re more effective in this popultion. Moreover, there re ongoing scientific debtes bout the effect of repeted vccintion on the immunologicl response induced by the sesonl influenz vccine [9-11] nd further evidence is needed. In 2011, we lunched pilot study to estimte the IVE ginst lbortory-confirmed influenz hospitlistion using network of hospitls in the Europen Union (EU) [12]. During the 2012/13 influenz seson, co-circultion of influenz A(H1N1)pdm09, A(H3N2) nd B/Victori- nd B/Ymgt-linege viruses ws reported in Europe [13]. The objective of the study presented here ws to mesure the 2012/13 sesonl IVE ginst hospitlistion with subtype-specific lbortory-confirmed influenz in hospitl network in four EU countries: Frnce, Itly, Lithuni nd Spin. Methods We conducted cse control study using the test-negtive design [14] in 18 hospitls locted in five study sites: Frnce (five hospitls), Itly (two), Lithuni (two), nd the Nvrre (four) nd Vlenci (five) regions in Spin. Ech study site dpted generic protocol [15] to the locl context (Tble 1). Study popultion The study popultion ws ll community-dwelling dults (18 yers of ge or older), belonging to the trget groups for vccintion s defined loclly [16-20], dmitted to one of the prticipting hospitls with no contrindiction for influenz vccintion. Ptients were excluded if they hd previously tested positive for influenz virus in the 2012/13 seson or resided outside the hospitl ctchment re (for the 11 hospitls with known ctchment re). Study tems ctively screened ll ptients dmitted for potentilly influenz-relted conditions. These conditions included the following: cute myocrdil infrction or cute coronry syndrome; hert filure; pneumoni nd influenz; chronic pulmonry obstructive disese; mylgi; ltered consciousness, convulsions, febrile-convulsions; respirtory bnormlity; shortness of breth; respirtory or chest symptoms; cute cerebrovsculr disese; sepsis; nd systemic inflmmtory response syndrome. Among them, study tems invited ptients with n onset of influenz-like 2

4 Tble 2 Definition of the ctegories of chronic conditions ccording to the vribles collected, hospitl-bsed influenz vccine effectiveness study, four Europen countries, 2012/13 Ctegories of chronic conditions Chronic conditions Study sites tht collected the informtion Crdiovsculr disese FR, IT, LT, VA Hert disese Crdiovsculr disese Stroke FR, IT, LT, NV Trnsient ischemic ttck IT Peripherl rteril disese IT, VA Lung diseses FR, IT, LT, NV Asthm IT, VA, LT Respirtory disese Chronic obstructive pulmonry disese IT, LT Emphysem IT, LT Mucoviscidosis FR, IT, LT Bronchitis VA, LT Dibetes FR, IT, NV, VA Metbolic nd endocrine disorders Nutritionl deficiency FR, IT, LT Endocrine disese FR, IT, LT, VA Hemtologicl cncer FR, IT, LT, NV Hemtologicl disese or cncer Anemi/spleen condition FR, IT, LT, VA Drepnocytosis FR, IT Cncer Immunodeficiency Immunodeficiency Rheumtologicl disese FR, IT, LT, NV Heptic disese Renl disese Obesity b Neuromusculr disorder FR, IT Dementi FR: Frnce; IT: Itly; LT: Lithuni; NV: Nvrre, Spin; VA: Vlenci, Spin. My include the conditions from the sme ctegory listed below. b Defined s body mss index 30 kg/m 2. illness (ILI) symptoms (one systemic nd one respirtory symptom) within the pst seven dys to prticipte. Those ccepting to prticipte were swbbed nd tested for influenz. Reverse trnscription polymerse chin rection (RT-PCR) ws used to detect influenz viruses nd to clssify them s influenz A(H3N2), influenz A(H1N1)pdm2009 or influenz B. Ptients positive for influenz were clssified s cses of given influenz type/subtype nd those testing negtive were controls. We defined the study period s t lest 15 dys fter the beginning of ech site-specific sesonl influenz vccintion cmpign until the end of the influenz seson s declred by locl influenz surveillnce systems. For ech of the influenz type/subtype nlyses, we excluded the controls with onset of symptoms before the week of the first lbortory-confirmed cse or fter the week of the lst lbortory-confirmed cse. We used the Interntionl Orgniztion for Stndrdiztion s week numbers [21] to ensure consistency cross study sites. We considered ptients s vccinted ginst sesonl influenz if they hd received t lest one dose of the 2012/13 influenz vccine more thn 14 dys before onset of ILI symptoms. Ptients not vccinted or vccinted less thn 15 dys before ILI onset were considered s unvccinted. Dt collection We collected dt on the ILI episode, demogrphics, chronic diseses (Tble 2), number of hospitlistions in the previous 12 months, number of consulttions t the generl prctitioner (GP) in the previous three months, smoking sttus, vccintion ginst influenz in 2012/13 nd 2011/12 nd, for those ged 65 yers nd over, functionl sttus before ILI onset using the Brthel score [22]. The dt were gthered from hospitl medicl records, fce-to-fce interviews with the ptient nd/or ptient s fmily nd lbortory dtbses. The vccintion sttus ws obtined from vccintion registers in two study sites, interview with the ptients nd/or ptient s fmily in two sites nd contct with the ptient s physicin in one site. 3

5 Tble 3 Number of records received by the pooled nlysis coordintor nd included in the pooled nlysis by study site, hospitlbsed influenz vccine effectiveness study, four Europen countries, 2012/13 Number of records per study site Type of record Nvrre, Vlenci, Frnce Itly Lithuni b Spin Spin Totl Eligible records ,535 2,329 Non-trget groups for vccintion Missing lbortory results Unknown vccintion sttus Totl records used for the nlyses ,390 1,972 Influenz A(H1N1)pdm09 Cses Controls ,213 1,513 Influenz A(H3N2) Cses Controls Influenz B Cses Controls ,294 In Frnce, one specimen of influenz A virus could not be subtyped. b In Lithuni, one ptient ws coinfected with A(H3N2) nd A(H1N1)pdm09 viruses. Dt nlysis Study sites trnsmitted nonymised dtsets to the pooled nlysis coordintor, through psswordsecured web-bsed pltform. We rn complete cse nlysis, excluding records for which lbortory results, vccintion sttus or potentil confounding vribles were missing. To test for heterogeneity between study sites, we used Cochrn s Q-test nd the I 2 index [23]. The Q-test provides p vlue tht indictes the presence or not of heterogeneity. The I 2 index quntifies the proportion of the vrince ttributble to differences between study sites. It is common to consider tht I 2 round 25%, 50% nd 75% indicte low, medium nd high heterogeneity, respectively. We conducted seprte nlyses for ech type/subtype of influenz. We estimted the pooled IVE s 1 minus the odds rtio (OR) (expressed s percentge) of being vccinted in cses versus controls, using one-stge method with study site s fixed effect in the model [24]. We ssessed the presence of effect modifiction by compring the time- nd study site-djusted OR (ssuming tht the test-negtive design cse control study is density cse control study implying djustment for the time of symptom onset) cross strt of chrcteristics using the homogeneity test. We considered vrible s confounder when the percentge chnge between the undjusted nd djusted OR ws greter thn 15%. We conducted multivrible logistic regression nlysis. In ddition to study site nd month of symptom onset, we djusted the models for the covrites identified s potentil confounders in the strtified nlysis s well s the presence of t lest one underlying condition nd the ge tht we modelled s restricted cubic spline with four knots [25]. The likelihood rtio test ws used to decide on the finl models. We conducted strtified nlyses by ge group (less thn 65 yers, yers nd 80 yers nd bove). To study the effect of previous influenz vccintion on lbortory-confirmed influenz, we conducted strtified nlysis using four vccintion sttus ctegories: vccintion in none of the sesons (2011/12 nd 2012/13), 2012/13 vccintion only, 2011/12 vccintion only nd vccintion in both sesons nd computed nd compred IVE for ech of these ctegories using vccintion in none of the sesons s reference. We crried out sensitivity nlyses excluding the weeks when less thn 10% of the ptients included were positive for influenz, excluding ptients who received ntivirls between the onset of symptoms nd swbbing nd by restricting the nlysis to ptients swbbed within four dys of symptoms onset. To void the inclusion of ptients with cute mnifesttion of chronic respirtory illnesses rther thn respirtory infection, we restricted our nlysis to ptients with no underlying respirtory conditions. We rn ll nlyses with Stt v12 (Stt Corp LP, College Sttion, TX, United Sttes). 4

6 Tble 4 Chrcteristics of influenz A(H1N1)pdm09 (n=116), influenz A(H3N2) (n=58) nd influenz B (n=232) cses nd corresponding test-negtive controls included in the study, hospitl-bsed influenz vccine effectiveness study, four Europen countries, 2012/13 (n=1,972) Chrctertistic Controls b (n=1,513) A(H1N1)pdm09 A(H3N2) B Cses (n=116) Controls c (n=519) Cses (n=58) Controls d (n=1,294) Cses (n=232) Number (%) e Number (%) e Number (%) e Number (%) e Number (%) e Number (%) e Medin ge in yers * Age group in yers (22.4) 60 (51.7)* 146 (28.1) 14 (24.1) 301 (23.3) 60 (25.9) (37.2) 42 (36.2)* 175 (33.7) 22 (37.9) 473 (36.6) 92 (39.7) (40.4) 14 (12.1)* 198 (38.2) 22 (37.9) 520 (40.2) 80 (34.5) Sex Mle 851 (56.2) 67 (57.8) 294 (56.6) 24 (41.4)* 718 (55.5) 108 (46.6)* Vccine sttus 2012/13 sesonl influenz vccintion 866 (57.2) 39 (33.6)* 296 (57.0) 20 (34.5)* 734 (56.7) 88 (37.9)* 2011/12 sesonl influenz vccintion 835 (55.3) 37 (31.9)* 296 (57.5) 25 (44.6) 702 (54.5) 102 (44.5)* Presence of comorbidities Metbolic nd endocrine disorders 546 (36.1) 41 (35.3) 195 (37.6) 24 (41.4) 462 (35.7) 72 (31.0) Crdiovsculr disese 768 (50.8) 49 (42.2) 247 (47.6) 26 (44.8) 636 (49.1) 103 (44.6) Renl disese 198 (13.1) 9 (7.8) 84 (16.2) 8 (13.8) 165 (12.8) 27 (11.7) Respirtory disese 750 (49.6) 50 (43.5) 243 (46.8) 25 (43.1) 634 (49.0) 80 (34.6)* Neuromusculr disorder 82 (5.6) 7 (8.0) 27 (5.9) 3 (6.4) 70 (5.7) 7 (3.7) Heptic disese 65 (4.3) 2 (1.7) 14 (2.7) 0 (0.0) 57 (4.4) 8 (3.5) Immunodeficiency 102 (6.7) 8 (6.9) 40 (7.7) 5 (8.6) 87 (6.7) 16 (6.9) Hemtologicl disese or cncer 321 (21.7) 16 (14.5) 96 (19.2) 12 (21.8) 279 (21.6) 30 (13.0)* Any chronic condition (of ll chronic conditions collected in the 1,404 (92.8) 106 (91.4) 473 (91.1) 52 (89.7) 1,195 (92.3) 192 (82.8)* study site) More thn one chronic condition 1,013 (67.0) 62 (53.4)* 340 (65.5) 37 (63.8) 853 (65.9) 113 (48.7)* Obesity f 423 (28.1) 26 (22.6) 127 (24.7) 10 (17.9) 359 (27.9) 54 (23.5) Pregnncy 10 (0.7) 1 (1.3) 7 (2.0) 0 (0.0) 11 (1.0) 8 (4.7)* Low functionl sttus g (mong ptients 65 yers) 232 (19.8) 9 (16.1) 5 (14.8) 4 (9.1) 187 (18.9) 34 (19.8) Other potentil confounders More thn one GP visit in previous 3 months 738 (49.1) 46 (39.7) 261 (51.3) 26 (46.4) 649 (50.7) 109 (48.0) Hospitlistions in previous 12 months 582 (38.5) 32 (27.6)* 205 (39.6) 22 (37.9) 502 (38.8) 70 (30.2)* Smoker sttus Current 277 (18.3) 39 (33.6)* 108 (20.8) 13 (22.4) 243 (18.8) 32 (13.9)* Former 580 (38.3) 35 (30.2)* 173 (33.4) 16 (27.6) 485 (37.5) 58 (25.1)* Never 656 (43.4) 42 (36.2)* 237 (45.8) 29 (50.0) 565 (43.7) 141 (61.0)* Potentil for misclssifiction Swbbing dely <4 dys 745 (49.2) 69 (59.5)* 233 (44.9) 24 (41.4) 621 (48.0) 90 (38.8)* Antivirl tretment before swbbing 18 (1.2) 12 (10.4)* 17 (3.3) 5 (8.6) 18 (1.4) 17 (7.3)* GP: generl prctitioner. * p vlue for difference between cses nd controls <0.05. Frnce, Itly, Lithuni nd Spin (Nvrre nd Vlenci regions). b Comprisons were mde with controls recruited between the week of the first cse of influenz A(H1N1)pdm09 nd the week of the lst cse of influenz A(H1N1)pdm09 (1,513 controls). c Comprisons were mde with controls recruited between the week of the first cse of influenz A(H3N2) nd the week of the lst cse of influenz A(H3N2) (519 controls). d Comprisons were mde with controls recruited between the week of the first cse of influenz B nd the week of the lst cse of influenz B (1,294 controls). e Unless otherwise indicted. f Defined s body mss index 30 kg/m 2. g Determined using the Brthel score [22]. 5

7 Tble 5 Influenz vccine effectiveness ginst influenz A(H1N1)pdm09, A(H3N2) nd B, djusted for vrious covribles by ge group, hospitl-bsed influenz vccine effectiveness study, four Europen countries, 2012/13 Groups ssessed A(H1N1)pdm09 A(H3N2) B All trget groups Number of cses nd controls 1, ,526 Number of cses; number of vccinted cses 116; 39 58; ; 88 Number of controls; number of vccinted controls 1,512; ; 296 1,294; 734 Vribles used for djustment of vccine effectiveness Percentge influenz vccine effectiveness (95% CI) Study site 47.0 (18.8 to 65.4) 54.4 (16.1 to 75.2) 46.5 (27.7 to 60.4) Study site nd month of symptom onset 45.7 (16.4 to 64.8) 53.0 (13.2 to 74.5) 44.3 (24.3 to 59.0) Study site, month of symptom onset nd ge 20.9 ( 25.3 to 50.1) 61.9 (27.2 to 80.1) 46.9 (26.8 to 61.5) Study site, month of symptom onset, ge nd presence of chronic conditions 21.3 ( 25.2 to 50.6) 61.8 (26.8 to 80.0) 43.1 (21.2 to 58.9) Ptients ged yers belonging to trget groups Number of cses nd controls 372 b 143 c 346 d Number of cses; number of vccinted cses 60; 9 14; 3 60; 7 Number of controls; number of vccinted controls 312; ; ; 91 Vribles used for djustment of vccine effectiveness Percentge influenz vccine effectiveness (95% CI) Study site nd month of onset 42.5 ( 28.3 to 74.3) 26.1 ( to 82.7) 68.4 (25.7 to 86.6) Study site, month of onset nd presence of chronic conditions 41.8 ( 30.7 to 74.1) NA c 66.0 (19.3 to 85.7) Ptients ged yers Number of cses nd controls 504 e 181 f 565 Number of cses; number of vccinted cses 42; 18 22; 7 92; 40 Number of controls; number of vccinted controls 462; ; ; 287 Vribles used for djustment of vccine effectiveness Percentge influenz vccine effectiveness (95% CI) Study site nd month of onset 44.2 ( 9.0 to 71.4) 55.7 ( 22.8 to 84.0) 37.3 ( 2.1 to 61.5) Study site, month of onset nd presence of chronic conditions 43.8 ( 10.7 to 71.5) 52.4 ( 33.9 to 83.1) 28.2 ( 18.9 to 56.6) Ptients ged yers Number of cses nd controls 623 g 216 h 600 Number of cses; number of vccinted cses 14; 12 22; 10 80; 41 Number of controls; number of vccinted controls 609; ; ; 348 Vribles used for djustment of vccine effectiveness Percentge influenz vccine effectiveness (95% CI) Study site nd month of symptom onset ( 1,170.7 to 41.9) 73.8 (30.0 to 90.2) 46.4 (9.6 to 68.2) Study site, month of symptom onset nd presence of chronic conditions NA g 73.8 (29.9 to 90.2) 44.8 (6.7 to 67.4) CI: confidence intervl; NA: not pplicble. Frnce, Itly, Lithuni nd Spin (Nvrre nd Vlenci regions). b A totl of 27 controls dropped becuse no cses in November mong ptients less thn 65 yers. c A totl of 17 controls dropped becuse no cses in December nd April nd in Itly mong ptients less thn 65 yers. No djustment for chronic disese becuse ll A(H3N2) cses ged less thn 65 yers hd chronic conditions. d A totl of 15 controls dropped becuse no cses in April mong ptients less thn 65 yers. e A totl of 101 controls dropped becuse no cses in December mong ptients ged yers. f A totl of 16 controls dropped becuse no cses in December nd in Nvrre, Spin, mong ptients ged yers. g Two controls dropped becuse no cses in Lithuni mong ptients ged 80 yers nd over. No djustment for chronic disese becuse ll A(H1N1)pdm09 cses ged 80 yers nd over hd chronic conditions. h Four controls dropped becuse no cses in April mong ptients ged 80 yers nd over. Results Overll, 2,329 eligible ptients, of whom 2,021 belonged to the trget groups for influenz vccintion, were recruited in the 18 study hospitls (Tble 3). A totl of 45 (2.2%) nd four (0.2%) ptients were excluded due to missing lbortory results nd missing vccintion sttus, respectively. We included totl of 1,972 ptients in the nlysis: 1,390 from Vlenci (177 cses), 350 from Frnce (121 cses), 87 from Lithuni (53 cses), 75 from Nvrre (28 cses) nd 70 from Itly (27 cses). Influenz A(H3N2), A(H1N1)pdm09 nd B co-circulted in ll study sites (Tble 1). The study site hving included ptients for the longest period of time ws Vlenci (week 47, 2012 to 15, 2013) nd for the shortest period ws in Itly (week 2 8, 2013). The period of 6

8 Tble 6 Crude nd djusted vccine effectiveness ginst influenz A(H1N1)pdm09 (n=1,625), A(H3N2) (n=571) nd B (n=1,518) by vccintion sttus, hospitl-bsed influenz vccine effectiveness study, four Europen countries, 2012/13 Influenz type Number of cses Number of controls Crude VE b (95% CI) Adjusted VE c (95% CI) A(H1N1)pdm09 (n=1,625) No vccintion in 2012/13 nd 2011/ /13 vccintion only ( 62.9 to 66.6) 6.2 ( to 58.2) 2011/12 vccintion only ( 47.1 to 75.4) 26.6 ( 81.6 to 70.3) 2011/12 nd 2012/13 vccintions (24.3 to 70.6) 27.9 ( 20.5 to 56.9) A(H3N2) (n=571) No vccintion in 2012/13 nd 2011/ /13 vccintion only ( to 95.7) 68.3 ( to 96.1) 2011/12 vccintion only ( to 64.9) 12.3 ( to 68.1) 2011/12 nd 2012/13 vccintions (1.7 to 73.8) 59.6 (18.5 to 80.0) B (n=1,518) No vccintion in 2012/13 nd 2011/ /13 vccintion only (27.6 to 87.2) 68.3 (24.5 to 86.7) 2011/12 vccintion only ( 73.1 to 42.7) 5.6 ( 84.5 to 39.6) 2011/12 nd 2012/13 vccintions (15.5 to 56.3) 37.3 (10.7 to 56.0) CI: confidence intervl; VE: vccine effectiveness. Frnce, Itly, Lithuni nd Spin (Nvrre nd Vlenci regions). b Adjustment for study site nd month of symptom onset. c Adjustment for study site, month of symptom onset, ge nd comorbidities. recruitment ws the longest for A(H1N1)pdm2009 (21 weeks) nd the shortest for A(H3N2) (15 weeks). Of the 1,972 ptients included in the pooled nlysis, 116 ptients tested positive for influenz A(H1N1) pdm09, 58 for A(H3N2) nd 232 for influenz B. Two ptients were coinfected with types A nd B nd one ptient ws coinfected with A(H3N2) nd A(H1N1) pdm09. One specimen of influenz A could not be subtyped. Influenz A(H1N1)pdm09 cses were younger (63 vs 77 yers, p<0.05) thn controls. A lower proportion of A(H1N1)pdm09 cses hd more thn one underlying condition (53.4% vs 67.0%, p<0.05), hd been hospitlised in the previous yer (27.6% vs 38.5%, p<0.05) nd higher proportion were current smokers (33.6% vs 18.3%, p<0.05) compred with controls (Tble 4). Influenz A(H3N2) cses nd controls were similr for ll chrcteristics except for the proportion of mle ptients (41.4% vs 56.6%, p<0.05). Compred with controls, lower proportion of influenz B cses hd underlying conditions (82.8% vs 92.3%, p<0.05), hd been hospitlised in the previous yer (30.2% vs 38.8%, p<0.05) nd were smokers (13.9% vs 18.8% of current smokers, p<0.05). The 2012/13 vccine coverge ws 57.2% mong ll controls (ll influenz-negtive ptients included in the study), 33.6% mong A(H1N1)pdm09, 34.5% mong A(H3N2) nd 37.9% mong influenz B cses (Tble 4). The p vlues ssocited with the Q-test nd the I 2 index using models djusted for ge, month of symptom onset nd chronic condition, testing for heterogeneity between study sites, were respectively 0.19 nd 40.0% for A(H3N2), 0.10 nd 48.3% for A(H1N1)pdm09 nd 0.08 nd 56.2% for influenz B. The overll djusted A(H1N1)pdm09 IVE ws 21.3% (95% confidence intervl (CI): 25.2 to 50.6; n=1,628); 41.8% (95% CI: 30.7 to 74.1; n=372) mong the yer-old ptients nd 43.8% (95% CI: 10.7 to 71.5; n=504) mong those ged yers. Among ptients ged 80 yers nd older, there were 14 A(H1N1)pdm09 cses, including 12 vccine filures (Tble 5). Restricted to those ged less thn 80 yers-old, the djusted IVE ws 35.2% (95% CI: 9.1 to 61.5; n=1,004). Adjusted IVE ginst A(H1N1)pdm09 ws 6.2% (95% CI: to 58.2; n=753) mong ptients vccinted in the 2012/13 seson only, 26.6% (95% CI: 81.6 to 70.3; n=724) for those vccinted in 2011/12 nd 27.9% (95% CI: 20.5 to 56.9; n=1,368) for those vccinted in both sesons (Tble 6). 7

9 Tble 7 Adjusted vccine effectiveness ginst influenz A(H3N2), influenz A(H1N1)pdm09 nd B viruses ccording to vrious restrictions, hospitl-bsed influenz vccine effectiveness study, four Europen countries b, 2012/13 Restriction Totl number/ number of cses A(H1N1)pdm09 A(H3N2) B Adjusted VE (95% CI) Totl number/ number of cses Adjusted VE (95% CI) Totl number/ number of cses Adjusted VE (95% CI) No restriction 1,628/ ( 25.2 to 50.6) 577/ (26.8 to 80.0) 1,526/ (21.2 to 58.9) No ntivirl tretment strted between symptom onset nd swbbing 1,598/ ( 30.7 to 49.3) 555/ (21.7 to 79.0) 1,491/ (17.3 to 57.2) Swbbing dely 4 dys 1,147/ ( 47.1 to 50.8) 359/ (10.0 to 82.5) 1,037/ (18.8 to 63.2) Weeks when rtio controls to cses ws <9:1 Ptients with no chronic respirtory conditions 1,019/ ( 15.1 to 57.2) 542/ (27.5 to 80.8) 1,142/ (21.6 to 60.4) 829/ ( 20.3 to 69.0) 304/ ( 4.3 to 82.9) 812/ (24.1 to 68.0) CI: confidence intervl: VE: vccine effectiveness. Adjustment for study site, month of symptom onset, presence of ny chronic condition nd ge. b Frnce, Itly, Lithuni nd Spin (Nvrre nd Vlenci regions). The overll djusted IVE ginst A(H3N2) ws 61.8% (95% CI: 26.8 to 80.0; n=577) (Tble 5). The djusted IVE ws 52.4% (95% CI: 33.9 to 83.1; n=181) mong yers ptients nd 73.8% (95% CI: 29.9 to 90.2; n=216) mong those 80 yers nd older. Among ptients ged less thn 65 yers, ll cses hd chronic conditions. In this ge group, the IVE djusted for month of symptom onset nd study site ws 26.1% (95% CI: to 82.7; n=143). Adjusted IVE ws 68.3% (95% CI: to 96.1; n=250) mong ptients vccinted in 2012/13 only nd 59.6% (95% CI: 18.5 to 80.0; n=488) mong ptients vccinted in 2011/12 nd 2012/13 (Tble 6). The overll djusted IVE ginst influenz B ws 43.1% (95% CI: 21.2 to 58.9; n=1,526), 28.2% (95% CI: 18.9 to 56.6; n=565) mong ptients ged yers nd 66.0% (95% CI: 19.3 to 85.7; n=346) mong those younger thn 65 yers (Tble 5). Adjusted IVE ginst influenz B ws 68.3% (95% CI: 24.5 to 86.7; n=714) mong ptients vccinted in 2012/13 only nd 37.3% (95% CI: 10.7 to 56.0; n=1,300) in those vccinted in both sesons (Tble 6). There were few chnges in the IVE when conducting the sensitivity nlyses (Tble 7). The IVE ginst A(H1N1) pdm09 ws higher when restricted to ptients with no chronic respirtory conditions (38.9% (95% CI: 20.3 to 69.0) vs 21.3% (95% CI: 25.2 to 50.6)). Discussion Our results suggest tht in the popultion trgeted for the influenz vccintion, the 2012/13 IVE for lbortory-confirmed hospitlised influenz ws 21.3% ginst A(H1N1)pdm09, 61.8% ginst A(H3N2) nd 43.1% ginst B. The dpttion of generic protocol by 18 Europen hospitls enbled us to pool dt nd obtin smple of 1,972 hospitlised ILI ptients trgeted for influenz vccintion. In seson with co-circultion of the three viruses, this lrge smple size llowed us to compute type-/subtype-specific estimtes of IVE ginst hospitlised influenz nd to further ttempt to strtify by ge group. However, strtified nlyses led to estimtes with brod confidence intervls. Consequently, some results of the strtified nlyses cn only be used to generte hypotheses. The test-negtive design hs been minly discussed nd vlidted for GP-bsed studies [26,27]. It is ssumed tht by restricting the study popultion to ptients consulting for ILI, the helth-seeking behviour confounding effect (ssocited with propensity to get vccinted nd to go to the GP in cse of influenz) is controlled for. Since in our study sites ll people needing hospitlistion re likely to be hospitlised, we believe tht confounding due to helth-seeking behviour is minimised. In hospitl-bsed studies, severl outcomes could be used. If we were to mesure IVE ginst influenz confirmed-severe cute respirtory infection (SARI), we would need to mke sure tht for both cses nd controls respirtory infection ws the cuse of dmission. We hve chosen broder cse definition nd more sensitive inclusion criteri to cover lrger prt of the influenz disese burden. As consequence, some of the ILI in the seven dys before dmission my correspond to n excerbtion of underlying respirtory conditions. This could led to n overestimtion of the 8

10 IVE. Restricting our nlysis to ptients with no underlying respirtory conditions provides similr results nd does not support this hypothesis. Furthermore, we djusted for the presence nd number of previous hospitlistions for underlying conditions. The inclusion of ptients swbbed more thn four dys fter symptoms onset or fter ntivirl tretment hd strted could hve led to misclssifiction bises if virl clernce occurred before swbbing. However, nlyses confined to ptients swbbed within four dys of symptom onset nd to ptients who did not receive ntivirl tretment did not chnge the results. Studies using the test-negtive design my underestimte the IVE when the rtio of controls to cses is high, especilly if the lbortory tests hve low specificity [28]. In our study, ll cses were confirmed by RT-PCR, which hs high specificity [29]. In the nlyses restricted to weeks when the control to cse rtio ws lower thn 9:1 resulted in very similr IVE estimtes. The dt qulity ws high with only 49/2,021 records with missing outcomes or exposures in the dtbse. We believe tht scertinment of vccintion sttus through ptient interviews in two of the five study sites hs not introduced differentil informtion bis s dt were collected before lbortory testing. Due to the smll smple size in some study sites, the test of heterogeneity my hve hd no power to detect heterogeneity even if differences exist between study sites. Different IVE cross study sites could be due to vritions in circulting strins, different vccines by study site or different mesured nd unmesured confounding fctors. Further typing of circulting strins would be vluble to discuss site-specific IVE with regrd to the level of mtching between vccine nd loclly circulting strins. Different ccess to vccintion ccording to ge nd underlying condition nd to hospitlistion [30] could prtly explin vritions in IVE cross study sites. Finlly, the presence of rndom errors cnnot be ruled out due to low smple size by study site. A lrger smple size would be needed to crry out two-stge pooled nlysis [24]. Our results suggest tht, in people belonging to trget groups for vccintion, the 2012/13 IVE vried by subtype nd ge group. However, we cnnot exclude the possibility tht the vribility of IVE results by ge group minly reflects smple size limittions. Smll strtum-specific smple sizes (nd very smll number of cses) led to unstble results nd do not llow for biologicl interprettion of ge-specific results. Our results would suggest tht IVE ginst A(H3N2) ws higher mong older ge groups. This observtion would be in contrdiction to the principles of immune senescence. In ddition to the smple-size limittions, nd s discussed bove, we cnnot exclude selection bis for our controls, which we djusted for. However we used the sme control group for the three subtypes nd ge-specific results vry by subtype. We consider tht it is unlikely tht confounding fctors would differ by subtype. When looking t the effect of repeted vccintion (over two consecutive sesons), similr ptterns were observed for influenz A(H3N2) nd B. The highest point estimte IVE ws in ptients vccinted in 2012/13 only, the lowest in those vccinted in 2011/12 only nd intermedite mong those vccinted both sesons. Such findings re consistent with recent reports from the Unites Sttes nd Austrli [9,10,31]. The 2011/12 vccine included n A/Perth/16/2009(H3N2)-like virus nd B/Brisbne/60/2008-like virus, while the 2012/13 vccine included n A/Victori/361/2011(H3N2)-like virus nd B/Wisconsin/1/2010-like (Ymgt linege). On the bsis of Europen virologicl surveillnce dt [13], the min circulting strins during the 2012/13 seson were n A/Victori/361/2011(H3N2) (with some A/Texs/50/2012 circultion reported) nd B/Wisconsin/1/2010-like (with some B/Estoni nd B/ Msschusets/2/2012 circultion reported). These dt support the bsence of protection by the 2011/12 sesonl vccine on the 2012/13 circulting strins s they were not mtched. Some uthors hve discussed the hypothesis of ttenuted immunologicl responses s result of repeted vccintion. From school-bsed study, Dvies et l. [32] suggested tht nturl infection in seson 1 produces ntibodies tht hve lrger potentil to form high post-vccintion titres in seson 2 thn vccineinduced ntibodies. Smith et l. [33] hypothesised tht lrge ntigenic distnces between vccines in sesons 1 nd 2, nd between vccine in seson 1 nd epidemic strin in seson 2, significntly increse the risk of infection mong repeted vccinees compred with those receiving the vccine in seson 2 only. Considering the ntigenic differences between the 2011/12 vccine nd the 2012/13 circulting strins, this hypothesis could explin our results, suggesting higher IVE ginst influenz A(H3N2) nd B mong ptients vccinted in 2012/13 only compred with those vccinted in 2011/12 nd 2012/13. Further studies, including longer history of vccine uptke nd nturl infections would be of gret vlue to better understnd the effect of repeted vccintion on the immunologicl response to new influenz sesonl vccine nd the level of clinicl protection conferred to individuls. Our results suggest low IVE ginst A(H1N1)pdm09, especilly mong the elderly [34]. A totl of 14 cses of influenz A(H1N1)pdm09 occurred mong ptients older thn 80 yers. While the mjority of these cses (n=12) were vccinted ptients, smll numbers mke the IVE estimtes hrd to interpret in tht ge group. The IVE ws similr for those vccinted in 2011/12 only or in both sesons. There ws no effect for those vccinted in 2012/13 only. The recommended A/Cliforni/7/2009(H1N1)pdm09-like virus strin 9

11 ws the sme for the 2011/12 nd 2012/13 vccines nd mtched the 2012/13 circulting strins (some A/Cliforni/06/2009 lso reported). Long-lsting immune response induced by trivlent inctivted vccines ws previously described [35] nd some recent results suggest tht frequent previous vccintions my be effective for the current influenz seson [11]. The bsence of protection mong ptients vccinted in 2012/13 only is difficult to understnd nd interpret; it my reflect the presence of ssocited (nd unmesured) negtive confounders for which repeted vccintion my be surrogte. In ddition, other studies [36-38] suggest decresing effect in the seson difficult to reconcile with long-term effect between sesons. Considering the smll smple size in some of the vccintion groups in our study, we cnnot exclude the possibility tht this observtion is due to chnce. Incresing the number of study sites in this network would llow sufficient smple size to be reched erly enough in the seson to prompt the use of lterntive prevention mesures if low IVE ginst hospitlised cses is observed mong the trget group. Erly estimtes of IVE ginst hospitlised influenz re lso useful complement to guide the decisionmking of WHO experts regrding the composition of the next seson s vccines. A lrger smple size nd good documenttion of vccine brnds used would llow the computing of brnd-specific IVE. To further study the effect of previous sesonl vccintion will require documenting pst vccintion over severl sesons. In ddition, wys to mesure pst nturl immunity my lso be needed to better understnd the complex immunity of influenz nturl infection nd vccintion. Acknowledgments We would like to thnk Vivek Shinde, Hélène Bricout, Clotilde El Guerche-Seblin, Bruno Cincio, Germine Hnquet nd Jim McMenmin for their scientific inputs in piloting this study. Mny thnks lso to EpiConcept collegues for their contributions: Thoms Seyler for inititing the network, Esther Kissling for her gret input on dt mngement, Mrt Vlencino for her reviews. We re grteful to ll ptients, medicl nd lbortory stff, study nurses nd epidemiologists from the four study sites who ctively prticipted in the study. Conflict of interest No conflict of interest. Snofi Psteur, GlxoSmithKline, Snofi Psteur MSD supported the study. They hd no role in study design, dt collection, pooled nlysis nd publiction. Authors contributions Mrc Rondy ws involved in the originl methodologicl design of the study (generic protocol). He coordinted the Europen hospitl IVE network, undertook the sttisticl nlysis on which the reserch rticle is bsed nd led the writing of the reserch rticle. Alin Moren initited the originl methodologicl design of the study. He coordinted the Europen hospitl IVE network nd contributed to the writing of the reserch rticle. Odile Luny, Jon Puig-Brberà, Giedre Gefenite, Jesús Cstill, Ktleen de Getno Donti, Florence Gltier, Eelko Hk, Mrcel Guevr nd Simon Costnzo were responsible for the coordintion of the study t the locl level. They were in chrge of the dt collection nd mngement. They red, contributed nd pproved the mnuscript finl version. The Europen hospitl IVE network contributors were in chrge of supervising the study t the hospitl level nd collected the dt published in this reserch rticle. Europen hospitl IVE network Frnce: Nezh Lenzi nd Zineb Lesieur, the French Clinicl Vccinology Network (Réseu Ntionl d Investigtion Clinique en Vccinologie REIVAC). Isbelle Bonmrin, Institut de veille snitire, Frnce. Xvier Duvl, Yolnde Cost, Annuxcy Kngrtnm, Yzdn Yzdpnh, Mrion Cseris, Nthlie Dournon, Thoms Ppo, Antoine Dossier, Hkim Bécheur, Anne-Lure Pelletier, Hervé Ml, Armelle Mrceu, Michel Aubier, Rphêl Bories, Enrique Cslino, Christophe Choquet nd Ndhir Houhou, CIC P-007, Hôpitl Bicht, Pris, Frnce. REIVAC. Pierre Loulergue, Reem Knn nd Florence Dums, Hôpitl Cochin, Pris. Inserm, CIC BT 505 Cochin Psteur, Pris, Frnce. REIVAC. Déborh Postil, Sébstien Alcolé nd Sylvie Rogez, CHU Dupuytren, Limoges. Inserm, CIC-P 1435, Limoges. REIVAC.Philippe Vnhems nd Corinne Régis, Groupement Hospitlier Edourd Herriot, Service d Hygiène, Epidémiologie et Prévention, Lyon. Université de Lyon 1, CNRS, UMR 5558, Lbortoire de Biométrie et Biologie Evolutive, Equipe Epidémiologie et Snté Publique, Lyon, Frnce. REIVAC. Corinne Merle, Vincent Foulongne, Mrine Ry, Véronique Mugueret-Doublet, Arnud Bourdin, Liline Lndreu, Amdou Konté, Philippe Corne, Mustph Sebbne, Kd Klouche nd Mrie-Suznne Léglise, CHRU Montpellier, Hôpitl Sint Eloi, Montpellier Frnce. REIVAC. Mrtine Vlette nd Bruno Lin, CNR des virus Influenz (Lyon) & lbortoire de Virologie Est des HCL, Groupement Hospitlier Est, BRON. Lbortoire Virpth, EA4610, Fculté de médecine Lyon EST, Université Lyon 1, Lyon. Fbrice Crrt nd Frédéric Chu, Epidémiologie des mldies infectieuses, Inserm UMR-S 707 Pris. Vlenci, Spin: Jvier Díez-Domingo nd Begoñ Escribno- López, Vlenci Hospitl Network: FISABIO, Vlenci. Alberto Arnedo-Pen nd Montserrt Ruiz-Grcí, Centro de Slud Públic de Cstellón, Cstellón:. Rmón Limón- Rmírez, Hospitl de l Pln, Vil-rel. Miguel Tortjd- Girbés, Hospitl Doctor Peset, Vlenci. Consuelo Crrtlá Munuer, Cátedr de Medicin de Fmili. Deprtmento de Medicin Clínic. Universidd Miguel Hernández, Sn Jun, Alicnte. Julio Brrenengo Sñudo nd Ros Lrre- González, Hospitl Generl, Cstellón:. Vicente Gil-Guillén, Hospitl de Eld, Alicnte. Germn Schwrz-Chvrri, Centro de Slud Sn Bls, Alicnte. Lithuni: Jnette Rhmt-Lngendoen, Hubert Niesters*, Deprtment of Medicl Microbiology, Division of Clinicl Virology, University of Groningen, the Netherlnds. Arvyds Ambrozitis nd Ligit Jncoriene, Deprtment of Infectious, Chest Diseses, Vilnius University, Lithuni. Aukse Mickiene, Monik Kuliese nd Div Velyvyte, Lithunin University of Helth Sciences, Deprtment of Infectious Disese, Kuns Clinicl Hospitl, Lithuni. Ronld P. Stolk, Deprtment of Epidemiology, University Medicl Centre Groningen, the Netherlnds. Kestutis Zgmins, Institute of Public Helth, Fculty of Medicine, Vilnius University, Lithuni: Nvrre, Spin: Crmen Ezpelet, Judith Chmorro nd Pilr Artjo, Complejo Hospitlrio de Nvrr, Pmplon. 10

12 Frncisco Lmeiro nd An Nvscués, Hospitl Grcí Orcoyen, Estell. Mite Orteg, Montse Torres nd José Jvier Grcí Irure, Hospitl Rein Sofí, Tudel. Fátim Irisrri, Mnuel Grcí Cenoz nd Iván Mrtínez-Bz, Instituto de Slud Públic de Nvrr, Pmplon. Itly: Roberto Cud, Concett Donto, Rosri Sntngelo, Rome, Ctholic University, Policlinico Gemelli. Frncesco Perlsc, Giovnni Ficher, Mrinn Dr, Vrese, Ospedle di Circolo e Fondzione Mcchi. Lici Icoviello, Pozzilli (IS), IRCCS Neuromed. Mrco Olivieri, Cmpobsso, EPICOMED Reserch, srl. * Authors correction At the request of the uthors, Hubert Niesters ws dded on 19 Jnury 2015 to the list of Europen hospitl IVE network members, Lithuni section. References 1. Crrt F, Flhult A. Influenz vccine: the chllenge of ntigenic drift. Vccine. 2007;25(39-40): Avilble from: PMID: Influenz vccines. Wkly Epidemiol Rec. 2005;80(33): Avilble from: PMID: Committee for Proprietry Medicinl Products (CPMP). Note for guidnce on hrmoniztion of requirements for influenz vccines. London: Europen Medicines Agency; Mr CPMP/BWP/214/96. Avilble from: em.europ.eu/docs/en_gb/document_librry/scientific_ guideline/2009/09/wc pdf 4. Coudeville L, Billeux F, Riche B, Megs F, Andre P, Ecochrd R. Reltionship between hemgglutintion-inhibiting ntibody titres nd clinicl protection ginst influenz: development nd ppliction of byesin rndom-effects model. BMC Med Res Methodol. 2010;10(1):18. Avilble from: org/ / PMID: McCullers JA, Huber VC. Correltes of vccine protection from influenz nd its complictions. Hum Vccin Immunother. 2012;8(1): Avilble from: hv PMID: World Helth Orgnition (WHO) Strtegic Advisory Group of Experts on immuniztion (SAGE). Bckground pper on influenz vccines nd immuniztion. SAGE Working Group. Genev: WHO; [Accessed 19 Apr 2013]. Avilble from: pril/1_bckground_pper_mr26_v13_clened.pdf 7. Mereckiene J, Cotter S, D Ancon F, Gimbi C, Nicoll A, Levy- Bruhl D, et l. VENICE project gtekeepers group. Differences in ntionl influenz vccintion policies cross the Europen Union, Norwy nd Icelnd Euro Surveill. 2010;15(44): Avilble from: PMID: World Helth Orgniztion (WHO). Recommended composition of influenz virus vccines for use in the northern hemisphere influenz seson. Februry Genev: WHO; Avilble from: virus/recommendtions/201202_recommendtion.pdf 9. Ohmit SE, Petrie JG, Mlosh RE, Cowling BJ, Thompson MG, Shy DK, et l. Influenz vccine effectiveness in the community nd the household. Clin Infect Dis. 2013;56(10): Avilble from: cid/cit060 PMID: Ohmit SE, Thompson MG, Petrie JG, Thker SN, Jckson ML, Belongi EA, et l. Influenz vccine effectiveness in the seson: protection ginst ech circulting virus nd the effect of prior vccintion on estimtes. Clin Infect Dis. 2014;58(3): doi: /cid/cit736 PMID: McLen H, Sundrm M, Kieke B, Meece J, McClure D, Thompson M, et l. Current nd previous seson vccintion hve similr effectiveness for preventing mediclly ttended influenz A (H3N2) nd B: results from community cohort over 8 sesons. Poster presenttion t Options for the control of Influenz VIII, Cpe Town, South Afric, 5-10 Sep Rondy M, Puig-Brber J, Luny O, Duvl X, Cstill J, Guevr M, et l sesonl influenz vccines effectiveness ginst confirmed A(H3N2) influenz hospitlistion: pooled nlysis from Europen network of hospitls. A pilot study. PLoS ONE. 2013;8(4):e Avilble from: org/ /journl.pone PMID: Europen Centre for Disese Prevention nd Control (ECDC). Influenz virus chrcteristion. Summry Europe, July Stockholm: ECDC; Avilble from: en/publictions/publictions/influenz-virus-chrcteristion- July-2013.pdf. 14. Vlencino M, Kissling E, Cincio BC, Moren A. Study designs for timely estimtion of influenz vccine effectiveness using Europen sentinel prctitioner networks. Vccine. 2010;28(46): Avilble from: org/ /j.vccine PMID: Seyler T, Rondy M, Vlencino M, Moren A. Protocol for hospitl-bsed cse control studies to mesure sesonl influenz vccine effectiveness ginst lbortory confirmed influenz hospitlistions cross the Europen Union nd Europen Economic Are Member Sttes. My I-MOVE. Avilble from : viewer?=v&pid=sites& srcid=zgvmyxvsdgrvbwfpbnxlcglmb HV8Z3g6NGY3ZGYyZDQ1OGQ3OGIxOA 16. Hut Conseil de l Snte Publique. Le Clendrier des vccintions et les recommndtions vccinles 2012 selon l vis du Hut Conseil de l snté publique. [2012 vccintion schedule nd recommendtions from the Hut conseil de l snté publique in Frnce]. Bulletin épidémiologique hebdomdire. 2012;14-15: French. Avilble from: Ministero dell Slute. Prevenzione e controllo dell influenz: rccomndzioni per l stgione [Prevention nd control of influenz: recommendtions for the seson ]. Rom: Ministero dell Slute; Itlin. Avilble from: ingu=itlino&menu=notizie&p=dlministero&id= Vccine Europen New Integrted Collbortion Effort (VENICE) II Consortium. Sesonl influenz vccintion survey in EU/ EEA, influenz seson Finl report. VENICE; Avilble from: Influenz_Vccintion_Survey_2010.pdf 19. Instituto de Slud Públic de Nvrr. Protocolo de vcuncion ntigripl , [ Influenz vccintion protocol]. Boletin informtive. 2012;69:1-4. Spnish. Avilble from: Vcuncion ntigripl estcionl [Influenz vccintion, 2012 seson]. Direccio Generl d Investigcio i Slut Public, Generlitt Vclencin. Spnish. 21. Interntionl Orgniztion for Stndrdiztion (ISO). ISO 8601:2004: Dt elements nd interchnge formts -- Informtion interchnge -- Representtion of dtes nd times. Genev: ISO; Mhoney FI, Brthel DW. Functionl evlution: the Brthel index. Md Stte Med J. 1965;14:61-5. Avilble from: PMID: Huedo-Medin TB, Sánchez-Mec J, Mrín-Mrtínez F, Botell J. Assessing heterogeneity in met-nlysis: Q sttistic or I2 index? Psychol Methods. 2006;11(2): Avilble from: PMID: Stukel TA, Demidenko E, Dykes J, Krgs MR. Twostge methods for the nlysis of pooled dt. Stt Med. 2001;20(14): Hrrell FE. Regression modeling strtegies: with pplictions to liner models, logistic regression, nd survivl nlysis. New York, NY: Springer; Fopp IM, Hber M, Ferdinnds JM, Shy DK. The cse testnegtive design for studies of the effectiveness of influenz vccine. Vccine. 2013;31(30): Avilble from: dx.doi.org/ /j.vccine PMID: Jckson ML, Nelson JC. The test-negtive design for estimting influenz vccine effectiveness. Vccine. 2013;31(17): Avilble from: vccine PMID: Orenstein EW, De Serres G, Hber MJ, Shy DK, Bridges CB, Grgiullo P, et l. Methodologic issues regrding the use of three observtionl study designs to ssess influenz vccine effectiveness. Int J Epidemiol. 2007;36(3): Avilble from: PMID: Petric M, Comnor L, Petti CA. Role of the lbortory in dignosis of influenz during sesonl epidemics nd potentil pndemics. J Infect Dis. 2006;194(s2) Suppl 2;S Avilble from: PMID: World Helth Orgniztion (WHO) Regionl Office for Europe. In-ptient cre dischrges per 100. Europen Helth for All dtbse (HFA-DB). Copenhgen: WHO Regionl Office for Europe. [Accessed 28 Dec 2014]. Avilble from: europen-helth-for-ll-dtbse-hf-db 11