Effectiveness of seasonal 2012 / 13 vaccine in preventing laboratory-confirmed influenza infection in primary care in the United Kingdom : mid-season analysis 2012 / 13

J McMenamin (jim.mcmenamin@nhs.net)1, N Andrews2, C Robertson1,3,4, D M Fleming5, H Durnall5, B von Wissmann1, J Ellis6, A Lackenby6, S Cottrell7, B Smyth8, M Zambon2, C Moore7, J M Watson2, R G Pebody2 1. Health Protection Scotland, Glasgow, United Kingdom 2. Health Protection Agency Health Protection Services – Colindale, London, United Kingdom 3. University of Strathclyde, Glasgow, United Kingdom 4. International Prevention Research Institute, Lyon, France 5. Royal College of General Practitioners Research and Surveillance Centre, Birmingham, United Kingdom 6. Health Protection Agency Microbiology Services – Colindale, London, United Kingdom 7. Public Health Wales, Cardiff, United Kingdom 8. Public Health Agency Northern Ireland, Belfast, United Kingdom


Background
In common with many countries the United Kingdom (UK) experienced unusually late influenza activity in 2011/12, with activity peaking only in week 8 of 2012 [1] in a season dominated by influenza A(H3N2) and characterised by excess in all-cause mortality and the occurrence of influenza outbreaks in nursing home settings [2,3].Trivalent seasonal influenza vaccine (TIV) provided only moderate initial protection against A(H3N2) infection in primary care in 2011/12 with subsequent significant intra-seasonal waning of protection [1].Overall in 2011/12 TIV VE against confirmed A(H3N2) infection, adjusted for age, surveillance scheme and month was 23% (95% confidence interval (CI): -10% to 47%).
These results, coupled with virological data on emerging strain types, supported the rationale for the change to the World Health Organization (WHO) recommendation for composition of the TIV for the northern hemisphere season for 2012/13 [4] in which the A/California/7/2009 (H1N1)pdm09-like virus was retained, and the A(H3N2) vaccine strain was updated to an A/Victoria/361/2011 (H3N2)-like virus.Additionally, the B/Victoria lineage influenza B vaccine component was replaced with an influenza B strain of the B/Yamagata lineage (B/Wisconsin/1/2010-like virus).Compared to season 2011/12, the UK has experienced a relatively early influenza season in 2012/13, which has been dominated by influenza B circulation, but also with some A(H3N2) circulation [5].The 2012/13 intra-seasonal estimation of vaccine uptake in individuals who are in clinical groups at increased clinical risk of complications [6] has indicated similar levels compared to the same time point in 2011/12 [5].The UK has an established surveillance scheme to produce interim and end of season estimation of the effectiveness of the influenza vaccine programme [1,7] and this paper presents the interim evaluation of the effectiveness of the 2012/13 vaccine.

Study population and period
Data were derived from five primary care influenza sentinel surveillance schemes in England (two schemes), Northern Ireland, Scotland and Wales.Details of the Royal College of General Practitioners (RCGP), Health Protection Agency (HPA) Specialist Microbiology Network (SMN), Public Health Wales, Public Health Agency of Northern Ireland and Health Protection Scotland (HPS) swabbing schemes have been presented previously [1,7].
The study period ran from 1 October 2012 to 4 January 2013.Cases were defined, as previously [1,7], as persons presenting during the study period in a participating general practitioner (GP) practice with an acute influenza-like illness (ILI) who were swabbed and then tested positive for influenza A or influenza B. ILI was defined as an individual presenting in primary care with an acute respiratory illness with either history of fever or documented temperature >38°C or complaint of feverishness [1,7].Patients were swabbed as part of clinical care, with verbal consent.Controls were individuals presenting with ILI in the same period that were swabbed and tested negative for influenza.
A standardised questionnaire collecting demographic, clinical and epidemiological information from cases and controls including date of birth, sex, defined underlying clinical risk group [1,7], date of onset of respiratory illness, date of specimen and influenza vaccination status for 2012/13 with vaccination dates was completed by the patient's responsible GP at the time of swabbing.

Laboratory methods
Laboratory confirmation was undertaken using realtime polymerase chain reaction (RT-PCR) assays for circulating influenza A viruses, influenza B viruses and other respiratory viruses [8,9].Samples in England were sent to the HPA Microbiology Services, Colindale (RCGP scheme) or one of the specialist HPA microbiology laboratories (SMN scheme).Samples in Wales were sent to the Public Health Wales Specialist Virology Centre and in Scotland to the West of Scotland Specialist Virology Centre (HPS scheme) for molecular testing.In Northern Ireland samples were sent to the Regional Virus Laboratory, Belfast.All participating UK laboratories are a designated WHO National Influenza Centre (NIC) and participate in WHO and UK quality assurance programmes.This aids RT-PCR assays to be comparable between laboratories.

Statistical methods
Persons with a date of onset between 1 October 2012 and 4 January 2013 were available for this analysis.If date of onset of symptoms was missing then the date the swab was taken was used to define the time of ILI.Persons were defined as vaccinated if the date of vaccination with the 2012/13 TIV was 14 or more days before onset of illness.Those in whom the period between vaccination and onset of illness was less than 14 days were excluded, as immunity is unknown.Patients were also excluded if the date of vaccination was missing, and samples with a delay greater than 29 days between onset of illness (where known) and sample collection were excluded as the sensitivity of the polymerase chain reaction (PCR) test reduces for long intervals between onset and sampling [10,11].
VE was estimated as 1-(odds ratio) using multivariable logistic regression models with influenza A or influenza B PCR results as outcomes and seasonal vaccination status as the linear predictor.In the analyses evaluating VE in preventing influenza A infection, samples positive for influenza B were excluded and vice versa.In the multivariable analysis the known confounders age (coded into five standard age groups, <5 years, 5-14 years, 15-44 years, 45-64 years and ≥65 years) and month of ILI onset (or swab taken if onset was unknown) were included as well as sex and surveillance scheme (HPS, RCGP, SMN, Northern Ireland, Wales).Effect modification by age and scheme was assessed by likelihood ratio tests.
All statistical analyses were carried out in Stata version 12 (StataCorp, College Station, Texas).

2012/13 influenza vaccine effectiveness
This report has information on 1,865 individuals from whom samples were collected during the study period.Of these, 957 samples were collected through the RCGP surveillance scheme, 293 through the SMN scheme, 511 through the HPS scheme, 41 through the Public Health Wales scheme and 63 in Northern Ireland.Table 1 shows the distribution and completeness of the baseline characteristics of the study participants according to whether they were cases or controls.
Those excluded from the study because of late swabbing, a time of less than 14 days between vaccination and onset of symptoms and missing information on vaccination are summarised in Table 2.There were therefore 1,324 persons (i.e.121 influenza A cases and 1,203 controls) for whom data on both vaccination status and influenza A infection were available.Similarly, there were 1,580 persons (i.e.377 influenza B cases and 1,203 controls) included in the estimation of trivalent vaccine for prevention of influenza B.

Vaccine effectiveness in prevention of influenza
Table 3 shows the number of samples positive and negative for influenza A, influenza B and the combined influenza A or B virus according to vaccination status.Crude and adjusted vaccine effectiveness are also shown.
The adjusted VE estimates (Table 3) were 49% (95% CI: -2% to 75%) for influenza A, 52% (95% CI: 23% to 70%) for influenza B and 51% (95% CI: 27% to 68) for influenza A and B combined.As seen in previous years, age and month of onset were associated with positivity and vaccination status and were therefore confounding variables.Risk group was missing for 158/1,865 (8.5%) and this variable was not included in the model as it was not significantly associated with swab positivity when added to the multivariable model and analyses in previous years had shown that this was not a confounding variable [1,7].Sex and surveillance scheme were retained in the model but did not change the VE estimates.When looking at effect modification there was no evidence that VE varied by scheme (p=0.26) or age (p=0.50).

Discussion
The early experience of the influenza season in the UK [5] and in a number of European Union (EU) Member States, the United States (US) and Canada [12] presents an opportunity for the generation of interim assessment of seasonal influenza vaccine effectiveness.This can be used to inform those countries yet to experience a significant season and to add to the evidence base around choice of vaccine composition for the northern hemisphere influenza season of 2013/14.Unlike the pattern in the current season in North America, where influenza activity has been dominated by influenza A(H3N2) followed by influenza B, the early influenza season experienced across the UK has been dominated by influenza B cases with noted homogeneity in this pattern of laboratory detection in each country (with the exception of the Scottish scheme in which influenza A(H3N2) cases appear to be in slight excess).Influenza B viruses from both the B/Yamagata Northern Ireland 40 No 923 (69) 315 (79) 98 (78) Yes 301 (22) 56 (14) 14 (11) Missing 116 (9) 28 (7) 14 (11) Interval between onset and sampling (days) and B/Victoria lineages have co-circulated during the 2012/13 influenza season in the UK, with the majority of influenza B isolates antigenically characterised to date belonging to the B/Yamagata lineage [5].This is the fifth season in which the UK pooled estimation of TIV VE has been undertaken [1,7].The data quality for the pooled analysis even at this interim analysis stage is deemed high with few missing data field entries.
This observational study of interim influenza VE for TIV against laboratory-confirmed influenza infection in primary care in the UK 2012/13 winter season, which would appear at this juncture to be a medium intensity influenza season with influenza B the dominant circulating strain, has two key findings: reassuringly the northern hemisphere 2012/13 TIV appears to offer moderate protection against the circulating influenza B strain; the point estimate for the TIV VE against influenza A is based on smaller numbers and, though not statistically significant at this stage, suggests a similar moderate level of protection.
These UK interim results which are adjusted for age, sex and calendar month within the season are consistent with the crude TIV VE reported in recent weeks in Morbidity and Mortality Weekly Report (MMWR) from this season's US experience to date [13] in which their season has been dominated by influenza A(H3N2) and with the overall adjusted VE estimate reported from Canada [14].Indeed the UK crude VE estimates prior to adjustment for age appear near identical to those in the US.
While the TIV VE for influenza B is statistically significant one should keep in mind that there is a reliance on a trivalent vaccine with only one influenza B component, which in any given influenza season may offer limited protection against another influenza B lineage not targeted by the vaccine [15].The availability of quadrivalent seasonal influenza vaccines licensed for use in the EU [16] would mean that this can be potentially averted.Work needs to be undertaken to demonstrate whether introduction of these vaccines would be cost-effective.
In conclusion, this study undertaken mid-season provides good evidence that this season's TIV provides protection against laboratory-confirmed influenza B infection and more limited evidence of likely protection against laboratory-confirmed influenza A infection The '-' sign indicates that the respective participant numbers were excluded from the vaccine effectiveness analysis.a When the symptom onset date was missing the date when the participant was swabbed was used.
b Numbers exclude the participants with missing vaccination history who additionally had an interval from symptom onset to sampling >29 days.c Numbers exclude the participants vaccinated 0-13 days before symptom onset who additionally had an interval from symptom onset to sampling >29 days and a missing vaccination history. in the patients attending their GP with influenza like illness in the UK.It is important to note that more precision in this estimate will be available at the end of the season, together with the ability to obtain agestratified estimates.Vaccination with the seasonal influenza vaccine remains the best protection against influenza.The results are consistent with a protective benefit from seasonal influenza vaccine.Within the UK and beyond, particularly in those countries who are either still early in their influenza season or who have evidence of continuing influenza transmission, it is important to stress that it is not too late to be vaccinated this season and individuals in clinical groups eligible for vaccination who have yet to be vaccinated should be encouraged to get vaccinated.
The 2012/13 influenza season in Canada has been characterised to date by early and moderately severe activity, dominated (90%) by the A(H3N2) subtype.Vaccine effectiveness (VE) was assessed in January 2013 by Canada's sentinel surveillance network using a test-negative case-control design.Interim adjusted-VE against medically attended laboratory-confirmed influenza A(H3N2) infection was 45% (95% CI: 13-66).
Influenza A(H3N2) viruses in Canada are similar to the vaccine, based on haemagglutination inhibition; however, antigenic site mutations are described in the haemagglutinin gene.

Background
The 2012/13 influenza season in North America has shown moderately severe activity, spiking over the December/January holiday period, with influenza A(H3N2) viruses predominating among typed/subtyped viruses to date in both Canada (about 90%) and the United States (US) (about 70%) [1,2].
The updated 2012/13 A(H3N2) reference strain recommended by the World Health Organization as vaccine component for the northern hemisphere (A/ Victoria/361/2011-like) is antigenically distinct from that recommended for the previous season (A/ Perth/16/2009-like) [3], with 11 amino acid (AA) residue differences at antigenic sites of the haemagglutinin (HA) surface protein [4].
Vaccine effectiveness (VE) in Canada was assessed by the country's sentinel surveillance network in January 2013.Here we report the interim 2012/13 VE estimates against the dominant circulating influenza A(H3N2) subtype in the context of antigenic and genetic characterisation of circulating strains.

Estimating influenza vaccine effectiveness
As previously described [5][6][7][8][9][10][11], a test-negative casecontrol design was used to estimate VE, whereby a patient presenting with influenza-like illness (ILI) testing positive for influenza virus was considered a case and a person testing negative was considered a control.
Several hundred community-based practitioners in sentinel surveillance sites across participating provinces (British Columbia, Alberta, Manitoba, Ontario and Quebec) may offer nasal or nasopharyngeal swabbing to any patient presenting within seven days of symptom onset of ILI -defined as acute onset of respiratory illness with fever and cough and one or more of the following: sore throat, arthralgia, myalgia or prostration.
The VE analysis period included specimens collected from 1 November 2012 (week 44: 28 October 2012-3 November 2012) to 23 January 2013 (week 4: 20-26 January 2013), taking into account onset of influenza activity (Figure 1) and an immunisation campaign that started in October.Epidemiological information was obtained from consenting patients or their parents/ guardians using a standard questionnaire at the time of specimen collection, before testing.Ethics review boards in each participating province approved this study.
Specimens were tested for influenza viruses A (to subtype level) and B at provincial reference laboratories by real-time reverse-transcription polymerase chain reaction according to provincial protocols [4,11].Odds ratios (OR) for influenza vaccination among cases versus controls were estimated by multivariable logistic regression.VE against medically attended laboratory-confirmed influenza was calculated as [1 -OR] × 100.Patients for whom the timing of vaccination was unknown or was less than two weeks before symptom onset were excluded from the primary VE analysis but explored in sensitivity analyses.Those with unknown comorbidity were included and further explored in sensitivity analyses.

Genetic characterisation of sentinel influenza A(H3N2) viruses
Sequencing of the HA1 gene of a convenience sample (n=82) of available influenza A(H3N2) viruses, spanning the season so far but with emphasis on more recent activity, was undertaken for each province to identify AA substitutions within the 131 residues of antigenic sites A-E [11,12].These were expressed as percentage identity and relatedness compared with the vaccine reference strain (A/Victoria/351/2011).Pairwise identities were calculated from alignments of translated protein sequences generated in Geneious Pro v4.8.5 using a MUSCLE multiple sequence alignment algorithm.The approximate likelihood method was used to generate the phylogenetic tree of aligned nucleotide sequences in Geneious Pro v4.8.5.
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).

Participants
A total of 939 specimens were submitted from sentinel surveillance sites between 1 November 2012 and 23 January 2013.After exclusion criteria were applied (Figure 2), 739 participants contributed to overall VE analysis: their profile was similar to that seen in VE Of 999 nasal or nasopharyngeal specimens collected between 1 October 2012 (week 40: 30 September−6 October 2012) and 23 January 2013 (week 4: 20−26 January 2013), we excluded from the epidemic curve specimens from the following patients: those failing to meet the influenza-like illness (ILI) case definition or for whom it was unknown (n=24); those whose specimens were collected more than seven days after symptom onset or for whom the interval was unknown (n=132); those whose age was unknown (n=1) and those for whom influenza test results were unavailable or indeterminate (n=9).Specimens were included regardless of the patient's vaccination status or timing of vaccination; specimens from patients with unknown comorbidity were also included.

Virus characterisation
All influenza A(H3N2) isolates to date this season characterised in Canada by the haemagglutination inhibition assay have been considered antigenically similar to the 2012/13 vaccine component, although characterisation so far includes few (n=3) of the sentinel viruses described here [1].HA1 sequences of a subset of 82 (29%) sentinel A(H3N2) viruses were thus assessed for substitutions potentially contributing to suboptimal VE (Figure 3, Table 5).Sequencing was based on original specimens from British Columbia (n=15), Alberta (n=25), Manitoba (n=4) and Ontario (n=11) and virus isolates from Quebec (n=27).
Of  3, Table 5).These Clade 6 viruses also included loss of glycosylation at position N45S, a non-antigenic site mutation.

Discussion
Mid  TIV: trivalent influenza vaccine.a Chronic medical conditions that place individuals at higher risk of serious complications (hospitalisation or death) from influenza as defined by Canada's National Advisory Committee on Immunization [13], including heart, pulmonary (including asthma), renal, metabolic (such as diabetes), blood, cancer, immune compromising conditions or those that compromise the management of respiratory secretions and increase the risk of aspiration or morbid obesity.Questionnaire was answered as 'yes', 'no' or 'unknown' to any one or more of these conditions without specifying.b Vaccine status was based on self/parental/guardian report.Detail related to special paediatric dosing requirements was not sought.
Immunised participants were predominantly offered split (non-adjuvanted) 2012/13 trivalent inactivated influenza vaccine during the regular autumn immunisation campaign.In British Columbia and Quebec, influenza vaccine is provided free of charge to high-risk groups [13].Others are encouraged to receive vaccine but must purchase it.In Ontario, Alberta and Manitoba, the vaccine is provided free of charge to all citizens aged ≥6 months.c In Canada, adjuvanted vaccine is approved for people aged ≥65 years and live-attenuated vaccine by nasal administration is approved for those aged 2-59 years [13]; their use, however, remains infrequent.Of the 47 people aged ≥65 years who were considered immunised in this study, 14 reported that they received adjuvanted vaccine and 19 did not know, while the rest would have received non-adjuvanted vaccine.Overall, 5/141 immunised participants and 5/18 immunised children aged ≤10 years reported intranasal administration.Vaccine effectiveness analysis was not stratified on that basis.d Immunised people who received the vaccine <2 weeks before symptom onset or for whom this was unknown were excluded from the primary vaccine effectiveness analysis.They were included for assessing 'any' immunisation regardless of timing and for comparison with other sources of vaccine coverage.The denominator is therefore shown for 'any' immunisation.e Children <2 years-old in 2012/13 were excluded from 2011/12 vaccine uptake analysis as they may not have been age-eligible in autumn 2011.
f Children <3 years-old in 2012/13 were excluded from 2010/11 vaccine uptake analysis as they may not have been age-eligible in autumn 2010.g Children <4 years-old in 2012/13 were excluded from influenza A(H1N1)pdm09 vaccine uptake analysis as they may not have been ageeligible in autumn 2009.
a Ontario was delayed while awaiting ethics board review, diminishing its contribution to this interim analysis.a Those with influenza A of unknown subtype were excluded from the A(H3N2)-specific analysis.b For the primary analysis, those with unknown comorbidity were coded as 'No' but explored in the sensitivity analysis as shown.c Those immunised <2 weeks before symptom onset or from whom a specimen was collected >7 days since symptom onset (or for whom these were unknown) were excluded but explored in the sensitivity analysis as shown.d Adjusted for age, comorbidity, province, interval, week.e Adjusted for age, province, interval, week.

Figure 3
Phylogenetic tree of influenza A(H3N2) viruses, Canada, 2012/13 sentinel surveillance system The phylogenetic tree was created by aligning the 82 Canadian sentinel sequences against sequences representative of emerging viral clades as described by the European Centre for Disease Prevention and Control (ECDC) [16] (n=10), A(H3N2) sequences collected globally between 1 November 2012 and 18 January 2013 (n=17), and recent vaccine strains (n=3).The global sequences were downloaded from Global Initiative on Sharing Avian Influenza Data (GISAID) by searching for human influenza A(H3N2) haemagglutinin sequences collected in the specified period (Table 1).
a Antigenic regions A-E comprise 131 amino acid residues [12].Only the 24 positions in those 131 residues showing mutations in the present study are displayed.British Columbia, Alberta, Manitoba and Ontario sequencing was performed on original specimens; Quebec performed the sequencing on virus isolates.b 2012/13 northern hemisphere vaccine reference strain (A/Victoria/361/2011) and other recent vaccine and variant reference strains.c 2012/13 northern hemisphere vaccine reference strain.d A total of 75 sentinel sequences clustered within Clade 3C, which also includes the 2012/13 A/Victoria/361/2011 vaccine strain ( [16] and Figure 3).Common to each of these 75 sentinel sequences however, were antigenic site mutations compared with the A/Victoria/361/2011 vaccine strain as shown in this table and summarised as follows, with the antigenic site shown in parentheses: Q156H (B), V186G (B), Y219S (D), N278K (C).Of these 75 sequences, 69 also showed N145S (A) while the other four included L157S (B).Of these 69 sequences, 14/22 Alberta and 2/4 Manitoba sequences additionally showed I67V (E) and 11/14 British Columbia, 1/4 Manitoba, 4/10 Ontario and 16/19 Quebec sequences included T128A causing loss of glycosylation site (B) as well as R142G (A) mutations.e Seven sequences clustered within Clade 6 (A/Iowa/19/2010-like; see [16] and Figure 3) with antigenic site mutations compared with the A/ Victoria/361/2011 vaccine strain as shown in this table and additional loss of glycosylation at non-antigenic site N45S (not shown).
our own adjusted VE estimates did not substantially differ (less than 5-10%) from our unadjusted VE estimates, assessment of bias and confounding has to be separately undertaken for each dataset.Nevertheless, suboptimal VE for the influenza A(H3N2) component of the vaccine in both Canada and the US is inconsistent with haemagglutination inhibition characterisation indicating good vaccine match to circulating A(H3N2) viruses [1,2].Such discordance between conventional in vitro characterisation of vaccine match by haemagglutination inhibition and epidemiological measures of VE has been noted in previous seasons' estimates from our sentinel network [6,7,11], highlighted also in a recent meta-analysis of other studies, including randomised controlled trials [18].
Molecular markers of virus mutation may offer more insight.It has previously been suggested that a change of at least four AA in two or more HA antigenic sites heralds emergence of virus drift, potentially compromising antibody binding [19].However, HA antigenicsite maps have been updated and more studies are needed to correlate genetic variation in circulating viruses with epidemiological variation in measured VE [12,20].Not only the number but also the nature and location of AA substitutions are likely to be relevant.Furthermore, hypotheses to explain the variable efficacy of repeat immunisation have included positive and negative interference from pre-existing antibody, with differential effects depending on the antigenic distance across successive vaccine components and circulating strains [21].We note that a high proportion of participants (91%) who were immunised this season had also received vaccine the previous season.These virological, host and other factors potentially contributing to suboptimal VE warrant more in-depth evaluation.
Limitations of this surveillance approach to VE estimation have been described previously [6][7][8][9][10][11].For our interim analysis, we draw particular attention to small sample size, resulting in wide confidence intervals and variability around the point estimate.Age-specific VE analyses (e.g.children and elderly people) would be of additional important interest -our estimates primarily reflect the prominent contribution of adults 20-49 years of age.However, stratification of VE analysis by age would further reduce the statistical power and precision of estimates in this interim report.The slightly higher VE with restriction to participants without comorbidity (Table 4) may similarly reflect such variability.End-of-season analysis will further expand upon these interim findings and may better support stratified analyses.

Introduction
Influenza is an important health problem that can lead to serious complications in persons with risk factors [1,2].Annual vaccination is the primary measure for preventing influenza and its complications [3].
Because the influenza vaccine composition is adapted each season to the viruses in circulation, its effectiveness varies [4].
Observational studies are the main way to evaluate vaccine effectiveness (VE) in each season, however, possible biases affecting comparability between vaccinated and unvaccinated persons must be overcome [5][6][7][8].Studies with non-specific outcomes tend to underestimate the VE [6], a problem that is resolved by analysing virologically confirmed cases [4,9].A design that compares confirmed influenza cases with testnegative controls ensures good comparability and is easy to carry out, thus this type of study has come to be widely used [4,9].
Song et al., in an immunogenicity study of the influenza vaccine, found that the antibody levels decline progressively, beginning in the first months after vaccine administration [10].In addition, people with higher risk of complications due to influenza may have a weaker immune response due to the immunodepression associated with some chronic diseases or to the immunosenescence associated with aging [11,12].
In Spain, influenza circulation in the 2011/12 season reached a peak in week 7 of 2012, the second latest peak in the past 15 seasons, after the 2005/06 influenza wave [13].Influenza A(H3N2) was the predominant virus in circulation, and a certain degree of vaccinevirus mismatch was observed [13].The objective of this study was to describe the effectiveness of the influenza vaccine in the 2011/12 season in preventing laboratory-confirmed influenza, including both outpatients and hospitalised patients.

Study population
The The seasonal vaccination campaign took place from 10 October to 25 November 2011.In Navarre the trivalent inactivated non-adjuvanted vaccine was recommended and offered free of charge to people aged 60 years or older and to those with risk factors or major chronic conditions [14].Other people can also be vaccinated if they pay for the vaccine.In the 2011/12 season, the vaccine included the strains A/California/07/2009(H1N1)-like, A/Perth/16/2009(H3N2)-like and B/Brisbane/60/2008like virus [15].Precise instructions for registering each dose were given to all vaccination points [14].
Influenza surveillance was based on automatic reporting of cases of influenza-like illness (ILI) from all primary healthcare centres and hospitals.Following the European Union case definition, ILI was considered to be the sudden onset of any general symptom (fever or feverishness, malaise, headache or myalgia) in addition to any respiratory symptom (cough, sore throat or shortness of breath) [16].A sentinel network composed of a representative sample of 76 primary healthcare physicians and paediatricians, covering 15% of the population, was asked to take nasopharyngeal and pharyngeal swabs, after obtaining verbal informed consent, from all their patients diagnosed with ILI whose symptoms had begun preferably less than five days previously.An agreed protocol of care for influenza cases was applied in hospitals, which specified early detection and nasopharyngeal swabbing of all hospitalised patients with ILI.Swabs were analysed by reverse transcription polymerase chain reaction (RT-PCR), and influenza-positive samples were subsequently typed/ subtyped as influenza A(H1 and H3), A(H1N1)pdm09 and B.About one in four positive swabs was randomly selected each week and sent to the National Influenza Centre-Madrid laboratory for genetic characterisation.

Study design and statistical analysis
We The cases were patients diagnosed with ILI in primary care or in hospitals who were confirmed for influenza virus by RT-PCR.The controls were patients with ILI in primary healthcare or in hospitals who were negative for influenza virus.Their vaccination status for the trivalent 2011/12 seasonal influenza vaccine was obtained from the online regional vaccination register [17].Subjects were considered to be protected starting 14 days after vaccine administration.From the electronic healthcare records we obtained the following baseline characteristics: sex, age, district of residence, migrant status (country of birth other than Spain has been related to a different pattern of use of healthcare services [18]), major chronic conditions (heart disease, lung disease, renal disease, cancer, diabetes mellitus, cirrhosis, dementia, stroke, immunodeficiency, rheumatic disease and body mass index of 40 kg/m2 or greater), hospitalisation in the previous 12 months and outpatient visits in the previous 12 months.

Figure
Vaccination status was compared between cases and controls.Different analyses were done: (i) comparing cases of each type of influenza with negative controls, (ii) including only patients in whom influenza vaccination was indicated because they were 60 years or older or had a major chronic condition, (iii) considering only patients in primary care, and (iv) including only swabs taken in the first four days after symptom onset.VE was also evaluated separately in two age strata (<65 and ≥65 years; a cut-off age different from that of the vaccination target population was chosen to match the one commonly used in similar studies), in two periods (weeks 50/2011 to week 8/2012 and weeks 9/2012 to 20/2012) and in three strata according to time since vaccination (<100, 100-119 and ≥120 days).
Percentages were compared by chi-square test.Logistic regression techniques were used to calculate the odds ratios (OR) with their 95% confidence intervals (CI).ORs were adjusted for potential confounders including healthcare setting, and for date of diagnosis grouped into four-week periods.VE was estimated as (1-OR)x100.

Description of cases and controls
The weekly number of swabbed patients followed the pattern of ILI incidence in the population (Figure).During the study period, 757 ILI patients were swabbed, 588 in primary healthcare and 169 in hospitals.Some 411 (54.3%) were confirmed for influenza virus: 382 for influenza A(H3), 28 for influenza B and one for influenza A(H1N1)pdm09.
Compared with confirmed cases of influenza, the group of test-negative controls had a higher proportion of males, of persons under the age of five years, people who had consulted a physician five or more times in the past year, who had major chronic conditions, and who were treated in hospital.There were no significant differences regarding migrant status or urban/ rural residence, and these variables were therefore not included in the multivariate analysis (Table 1).Vaccine coverage in controls (18.8%) was slightly higher than in the overall population cohort in which the study was nested (14.2%, p=0.015).

Effectiveness of the 2011/12 seasonal influenza vaccine
Compared with test-negative controls, a smaller proportion of confirmed influenza cases had received the 2011/12 seasonal influenza vaccine (OR: 0.60; 95% CI: 0.40 to 0.89; p=0.012).In the adjusted analysis, the VE was 31% (95% CI: -21 to 60; p=0.200).The VE was somewhat higher between weeks 50/2011 and 8/2012 In none of all the analyses was the VE statistically significant (Table 2).
The VE was 61% (95% CI: 5 to 84) in the first 100 days after vaccination, dropping to 42% (95% CI: -39 to 75) between days 100 and 119, and ceasing to confer any protection after 120 days (-35%, 95% CI: -211 to 41) (Table 3).Persons vaccinated more than 120 days before diagnosis versus those vaccinated less than 100 days before diagnosis were at an increased risk for contracting influenza, with an OR of 3.45 (95% CI: 1.10 to 10.85; p=0.034).The point estimates of the influenza VE ranged between 61% and 69% during the first 100 days after vaccination in the analyses restricted to cases of influenza A(H3), to persons with an indication for influenza vaccination, to primary care patients, and to hospitalised patients, although this last result was not statistically significant.However, in all these analyses the vaccine had practically zero effectiveness at 120 or more days after vaccination (Table 3).In persons under 65 years of age the VE declined little with time since vaccination, whereas in those aged 65 years or older the OR for the risk of influenza was 20.81 (95% CI: 2.14 to 202.71; p=0.009) for those vaccinated more than 120 days previously versus those vaccinated less than 100 days previously.

Genetic characterisation
In total 102 isolates obtained from the confirmed cases, were further characterised by phylogenetic analysis of the HA1 sequence of the haemagglutinin gene in the National Influenza Centre -Madrid laboratory: 90 were influenza A(H3N2), 11 influenza B and one was influenza A(H1N1)pdm09.The strains most frequently identified were similar to A/Victoria/361/2011(H3N2) (

Discussion
The results of this study suggest that on average, the seasonal influenza vaccine had a low protective effect in preventing laboratory-confirmed influenza during the 2011/12 season in Navarre.Most of the strains we characterised showed reduced reactivity with postinfection ferret antiserum raised against the vaccine viruses, suggesting a certain degree of vaccine-virus mismatch [19].Although the confidence intervals were wide, similar estimates were obtained in analyses restricted to the target population for vaccination, to primary healthcare patients, or to patients swabbed within the first four days after symptom onset, which strengthens the conclusion and rules out possible biases.Evaluation of VE in preventing cases of influenza A(H3) only, also yielded similar estimates.
The early estimates of influenza VE for the first part of the season were higher than what we found for the complete season [20,21], which suggests a decline in VE over time.Two possible mechanisms, or a combination of both, could explain this reduced VE.The first is a change in the viruses circulating during the season, either due to appearance of another virus type or due to antigenic drift of circulating viruses, resulting in a loss of the match with the vaccine viruses.Our results do not support this mechanism, since the only relevant change in the circulating viruses was an increase in influenza B viruses, and low VE was also observed when we evaluated the effectiveness of the vaccine against influenza A(H3) only.
The second possible mechanism is waning immunity in those who received the vaccine.It has been reported that antibody levels begin to fall one month after administration of the influenza vaccine [10].This loss of immune response is more pronounced in older persons [10][11][12].The results of our study show a decline in the VE beginning 100 days after vaccination, primarily in persons aged 65 years or older.This finding could be explained by an immunosenescence phenomenon, aggravated by the long time between vaccination and virus circulation, which was longer than in most other seasons [13], and the limited match between vaccine and circulating strains [20,21].
Longer time between symptom onset and swabbing has been associated with reduced sensitivity in virus detection, which could underestimate VE [6].We controlled for this effect mainly in the design of our study, since 99% of the swabs from primary healthcare patients were taken within the first four days after symptom onset.Moreover, we repeated the analysis after eliminating the cases swabbed after the first four days, and no relevant changes in the estimate of VE were found.
The present study included both outpatient and hospital cases systematically recruited in a previously defined population.Primary care patients made up the bulk of subjects in our study and, when the analysis was limited to these patients, the VE was maintained.
The number of cases treated in hospitals was small, which did not allow us to obtain a specific estimate of the VE in preventing hospitalised cases.
Although institutionalised patients were not included in this study, several influenza outbreaks in nursing homes with high vaccination coverage were detected in Navarre in the 2011/12 season [22].This may be considered another consequence of the low VE.
This case-control analysis included only laboratoryconfirmed cases and compared them with test-negative controls recruited in the same healthcare settings before either patient or physician knew the laboratory result, a fact that provides better comparability and reduces selection bias [6].This type of design has been used in other studies that have evaluated influenza VE [20,21,23,24].The case¬-control study was nested in a population cohort for which extensive and reliable databases were available, and which was treated in hospitals and primary healthcare by physicians trained to detect and swab ILI patients, all of which can prevent unmeasured confounding [25].
In interpreting the results, some limitations must be kept in mind.The study size was insufficient to demonstrate a VE under 40%, which was reflected in wide confidence intervals that included zero.It may not be possible to generalise the results and apply them to other geographical areas, although other published studies in the same influenza season are consistent with our data [20,21].Although RT-PCR has high sensitivity for the detection of influenza virus, we cannot completely rule out some false negative results, which would cause a small underestimation of the VE.Although all the analyses were adjusted for the commonly recognised confounding factors, some residual confusion is possible [6].
These results suggest that VE may vary throughout the influenza season.The early estimates of influenza VE obtained in mid-season may drop during the season.This situation should be kept in mind given its implications for clinical practice and public health; it should not be interpreted as an error in the estimates, but as a description of reality.These early estimates remain enormously useful in redirecting preventive strategies during the influenza season and because they can aid the selection of strains to be included in the following season's vaccine [20,21].
The description of situations in which influenza VE is low should serve as a stimulus to design better influenza vaccines [26], to improve the selection of strains contained in the vaccine, to choose the most appropriate time for vaccination in each area, to encourage vaccination of caregivers of high-risk individuals, and to highlight the importance of other preventive measures that complement vaccination in high-risk populations, such as promotion of basic hygiene measures and avoidance of contact with influenza cases [27].
Early treatment with antiviral drugs should be considered in persons diagnosed with influenza who have a high risk of complications, regardless of vaccination status [28].In seasons when influenza starts late, it may be useful to revaccinate persons with a high risk of complications, especially those who may have a reduced immune response due to immunosenescence or immunodepression.
Even in seasons in which the effectiveness of influenza vaccine is low, it may appreciably reduce the number of cases and hospitalisations in high-risk persons.In the 2011/12 season in Navarre, the vaccine managed to avoid almost one third of the influenza cases in the vaccinated at-risk population; while not entirely satisfactory, this result is important in terms of individual and public health.

Conclusions
Our

Laboratory methods
Laboratory confirmation was undertaken using realtime polymerase chain reaction (RT-PCR) assays for circulating influenza A viruses, influenza B viruses and other respiratory viruses [6,7].Samples in England were sent to the HPA Microbiology Services, Colindale (RCGP scheme) or one of the specialist HPA microbiology laboratories (SMN scheme).Samples in Wales were sent to the Public Health Wales Specialist Virology Centre and in Scotland to the West of Scotland Specialist Virology Centre (HPS scheme) for molecular testing.
In Northern Ireland samples were sent to the Regional Virus Laboratory, Belfast.Influenza viruses were isolated in MDCK or MDCK-SIAT1 cells from RT-PCR positive samples as previously described [8].Virus isolates were characterised antigenically using post-infection ferret antisera in haemagglutination inhibition (HI) assays, with guinea pig red blood cells [9]

Statistical methods
Persons were defined as vaccinated if date of vaccination with the 2011/12 TIV was 14 or more days before onset of illness.Those in whom the period between vaccination and onset of illness was less than 14 days were excluded, as their immune status was unclear.
If the date of vaccination was missing, as the 2011/12 campaign occurred before influenza circulation, it was assumed that TIV vaccination was more than14 days before onset date.If date of onset of symptoms was missing then the date was assumed to have been four days before the swab was taken (the median interval based on the observed data).Respiratory samples with a delay greater than 29 days between onset of illness and sample collection were excluded as the sensitivity of the PCR test decreases for long intervals between onset and sampling.A sensitivity analysis was also undertaken, censoring at seven days between onset of illness and sample collection.
VE was estimated as 1-(odds ratio) using multivariable logistic regression models with influenza A(H3N2) or influenza B PCR results as outcomes and seasonal vaccination status as the linear predictor.In the analyses evaluating VE in preventing influenza A(H3N2) infection, samples positive for influenza B were excluded, and vice versa.Age (coded into five standard age groups, <5 years, 5-14 years, 15-44 years, 45-64 years and ≥65 years), sex, clinical risk group, surveillance

Table 1
Inclusion and exclusion criteria of participants for specimens submitted, United Kingdom, October 2011-April 2012 Criteria N Excluded N Included

Original participants 3,869
Excluded as interval from onset to sampling >29 days 81 Remaining participants 3,788

Analysis of TIV 2010/11
Excluded as missing vaccination history 166 Excluded as vaccinated 0-14 before onset 62

Final remaining study participants 3,560
Final for assessment of influenza A(H3N2) 3,517 Final for assessment of influenza B 3,184 TIV: trivalent seasonal influenza vaccine.

Results
A total of 3,893 individuals were swabbed in primary care during the study period.Six were excluded because they were positive for influenza A(H1N1) pdm09, two because the swab result was inconclusive and 16 because no laboratory result was available.This left 3,869 persons in the analysis.Table 1 summarises which of those individuals were excluded from the analysis of effectiveness.
Of these 3,869, 2,038 (52%) were collected from the RCGP scheme, 1,296 (33%) from the HPS scheme, 348 (9%) from the SMN scheme, 61 (2%) from the Public Health Wales Scheme and 126 (3%) from the Northern Ireland Scheme.The demographic and epidemiological characteristics of cases and controls are summarised in Table 2.There were statistically significant differences between cases and controls for all variables in Table 2. Vaccine date was unknown for 148 individuals who had received TIV.Although date of onset was missing for 263 (7%) individuals, these were included with onset date defined as swab date minus four days.

Model fitting for vaccine effectiveness estimation
When estimating vaccine effects, age group, sex, time period (defined by month of sample collection) and surveillance scheme were adjusted for in a multivariable logistic regression model.Although all these variables were significantly associated with having a positive swab, only age group and month of sample collection were confounders for the vaccine effects.Tables 3, 4 and 5 show vaccine effectiveness estimates against influenza A(H3N2) and B according to vaccination status and time since vaccination and period.
The adjusted age-specific estimates suggested protection was lower in the middle age groups (15 to 64 years), although the observed differences were not significant.There were significant differences in VE in relation to the interval since vaccination, with an adjusted VE of 53% (95% CI: 0 to 78) if the time from onset to vaccination was less than three months, compared with 12% (95% CI: -31 to 41) if the time was three months or more (test for trend: p=0.02).
The adjusted VE for TIV 2011/12 against influenza A(H3N2) with time since vaccination and interval from onset to swab included in the model is shown in Table 5.There was no significant difference in adjusted VE by scheme or by time from onset to swab (Table 5).Information on risk group was missing for 395 of 3,869 samples (10.2%) and was therefore not included in the final model.If risk group was included, the VE estimates remained unchanged.

Vaccine effectiveness against influenza B infection
The adjusted VE of TIV against influenza B was 92% (95% CI: 38 to 99) adjusted for age group, sex, time period and surveillance scheme.There no evidence that the VE varied by age group, although the numbers were small (with only a single vaccinated influenza B case with a B/Yamagata lineage infection).It was therefore not possible to stratify by time since vaccination, or by time period, to determine if there was reduction in protection.

Antigenic characterisation of circulating A(H3N2) viruses
The b Note that positive swabs from SMN and HPS were mainly taken after January 2012 with only four and six positive samples by January, respectively.RCGP had 65 positive swabs by January and gave a VE estimate for samples up to January of 50% (95% CI: -25 to 80), and one of 59% (95% CI: 1 to 83) for those vaccinated within three months before symptom onset.c Test for trend using time since vaccination as continuous.raised against influenza A/Perth/16/2009 (fourfold difference in HI assays; Table 6).A more than fourfold difference in HI assay titres with reference antiserum is considered to be significant antigenic drift [10].The proportion with a fourfold difference increased but did not change significantly over the duration of the 2011/12 season (from 13% in the period October 2011 to January 2012 to 21.9% in the period February 2012 to April 2012).Antigenic analysis of A(H3N2) virus isolates from combined sentinel and non-sentinel sources, confirmed the change in proportion over the two time periods to be non-significant (data not shown).

Discussion
This observational study of influenza VE for TIV against laboratory-confirmed influenza infection in primary care in the UK 2011/12 winter season, a late, low intensity influenza season with A(H3N2) as the dominant circulating strain, has several key findings: firstly, the 2011/12 seasonal influenza vaccine was overall poorly protective in preventing influenza A(H3N2) infection; secondly, vaccine protection was moderate in the first three months of the season, but reduced in the second three months; thirdly, there was evidence of waning protection against influenza A(H3N2) three months after vaccination; and finally, the 2011/12 TIV was highly protective against the circulating influenza B strain.
The test-negative case-control study design is becoming an increasingly well established approach to measure influenza vaccine effectiveness [11,12].One criticism of the method relates to the selected control population (test-negatives).In fact, use of this control group of individuals consulting in primary care with a respiratory illness that is not influenza is believed to overcome differences in health-seeking behaviour between cases and controls.Another criticism relates to the inclusion of individuals who were tested up to 29 days after disease onset, rather than those tested within seven days of onset.It is argued that test sensitivity declines with time from onset to swab and that such an approach may result in misclassification of cases as controls.We demonstrated that restricting samples to those taken within seven days of symptom onset did not significantly change the estimated vaccine effectiveness, although it did lead to loss of power as individuals were discarded.We did not adjust for multiple testing because waning was a priori of interest and was an objective of the study.This study based on surveillance data only had access to limited information on confounders.However, observational VE studies based on routine electronic health data in primary care using RCGP data [13] suggest that the most important confounders have been captured in our analysis.Indeed in our paper, we found risk status was not an important confounding variable, and to maximise power it was not included in the final multivariable analysis.
Our study demonstrates that during the 2011/12 influenza season, the 2011/12 TIV was overall poorly effective (with a non-significant adjusted VE of 23%) in protecting against confirmed influenza A(H3N2) infection for persons consulting their general practitioner (GP) with an ILI.Early estimates from the 2011/12 season have been published by several other countries -including a pooled case-control study from several European countries [3] and a study from Spain [4], demonstrating a low to moderate VE (43% and 55% respectively).It has been postulated in these studies that this could be due to a combination of a poor match between the 2011/12 TIV A(H3N2) virus strain (A/ Perth/16/2009) and the circulating A(H3N2) virus, and a waning protection.In the UK we found that the majority of characterised A(H3N2) viruses were antigenically similar to the vaccine component, with a notable proportion of A(H3N2) viruses showing some reduced reactivity in antigenic characterisation assays, but no significant change in that proportion over the duration of the 2011/12 season.Thus a certain degree of mismatch may explain the initial moderate protection, but does not seem to provide a complete explanation for the observed reduction in vaccine effectiveness over the course of the season and with increasing time since vaccination.These observations could challenge our current view on how mismatch is to be defined -an issue highlighted by Skowronski et al. [14] An alternative explanation may be waning immunity.
Our study demonstrates that influenza A(H3N2) vaccine effectiveness was higher in the first three months of the 2011/12 season compared to the last three months.In addition, TIV VE was moderate and significantly higher when disease onset was within three months of vaccination compared to three months or more.The UK, indeed, experienced an extremely late and mild influenza season in 2011/12, with influenza A(H3N2) activity not peaking until week 8 in 2012, such as has rarely been observed in previous GP weekly consultation data from RCGP (for example activity peaked in week 11 in 1993 when the dominant circulating strains were A(H1N1) and B, with both strains included in the vaccine).This present observation was accompanied by reports of outbreaks of influenza A(H3N2) in nursing home settings, which frequently had a high proportion of vaccinated persons [1].Waning intraseasonal vaccine protection would provide an explanation for these observations.At least two published studies have demonstrated intraseasonal waning in antibody titre following seasonal influenza vaccination [15,16].Bothshowed a significant reduction in antibody titre in elderly populations 20 to 22 weeks after vaccination.This would provide a biological explanation for our observed reduction in vaccine effectiveness over this particularly late season, where the median time from vaccination to disease onset was approximately three months.There are few reports of this in the literature: a large summertime outbreak due to circulation of a drifted A/Sydney/05/97-like (H3N2) virus reported in elderly tourists in Alaska was reported to have been due to a combination of drift and waning immunity [17].Our study was not adequately powered to be able to examine age-specific differences in waning and to determine if the effect was particularly marked in the elderly.
The 2011/12 TIV VE estimate against influenza B demonstrates high protection.This corresponds only partially with the virological data, which shows that in 2011/12, both B/Yamagata-lineage and B/Victoria-lineage influenza B viruses co-circulated in the UK.Furthermore the majority of influenza B circulated late in the season, like the A(H3N2) virus [1].Thus although we were not able to formally examine if there had been a reduction in protection connected to either time in the season or time since vaccination, effectiveness against influenza B was still high at the end of the season, with single vaccine failure occurring in a person infected with the B/Yamagata-lineage non-vaccine strain.
In conclusion, this end of season study provides important evidence that the 2011/12 season's TIV provided good protection against influenza B, but overall poor protection against the dominant circulating influenza A(H3N2) virus.This observation seems to be at least partially related to waning protection.The relative contributions of waning immunity and vaccine mismatch are unclear.This highlights the importance of future work to examine this phenomenon further.The study, however, reinforces the recommendation that annual re-immunisation of target groups is required regardless of TIV vaccination the previous season.The concept that vaccine protection can be so short-lived provides a challenge for public health policy.Influenza immunisations are given before the start of the influenza season when vaccine becomes available.In many winters, protection will therefore be optimal when the peak period of activity occurs in the first half of the winter.Influenza activity, however, can occur in the second half of the winter season, when protection may be waning.This highlights the pressing need for the development of influenza vaccines which provide better and longer-lasting protection, whether in terms of antigen content or formulation, e.g. through the use of adjuvants.In the interval, until such vaccines become available, this poses a policy question about whether there is a role for a second dose of seasonal influenza vaccine in certain circumstances: for example, when faced with late season outbreaks particularly in the groups most at risk of complications.
Our findings reinforce the need for annual revaccination and for early intraseasonal estimates of vaccine effectiveness to provide information for public health action, in particular to inform the annual WHO recommendation for composition of the vaccine for the following season.The identification of low or moderate vaccine effectiveness may allow communication of public health messages to clinicians to suspect influenza infection even in their highly vaccinated populations and have a lower threshold for prescribing of antiviral drugs to prevent the worst complications of influenza.

Introduction
Unlike the formulation of other vaccines, the formulation of seasonal influenza vaccines is reviewed annually by the World Health Organization (WHO) and frequently adapted to the constantly evolving nature of influenza viruses.
How the vaccine performs in target group populations cannot be anticipated by pre-authorisation efficacy trials in healthy young adults, immunogenicity studies or the relatedness of vaccine and circulating viruses.Field influenza vaccine effectiveness (IVE) studies provide essential additional information to advise stakeholders on the performance of the vaccine, to contribute to vaccine strain selection process and to inform when additional measures, such as antivirals, are needed given a low observed effectiveness early in the season.
In the European Union (EU) countries, the seasonal influenza vaccine is recommended annually for specific target groups, including those at risk of severe disease, the largest groups being older individuals (generally 60 or 65 years and above, depending on the country) and all those over six months of age with underlying medical conditions in the following categories: chronic respiratory and cardiovascular diseases, chronic metabolic disorders, chronic renal and hepatic diseases and immune system dysfunctions (congenital or acquired) [1].

Methods
The eight study sites included in the 2011/12 I-MOVE multicentre case-control study were based in France, Hungary, Ireland, Italy, Poland, Portugal, Romania and Spain.At each study site, practitioners already participating in the European Influenza Surveillance Network (EISN) were invited to take part in the study [19].In addition, study sites in Hungary and Portugal invited practitioners outside the EISN network.
The study population consisted of non-institutionalised influenza-like-illness (ILI) patients without contraindications for vaccination who were swabbed within less than eight days after symptom onset.Practitioners carried out naso-pharyngeal swabbing and collected information from patients consulting for ILI or, for France only, for acute respiratory infection (ARI).Only patients adhering to the EU ILI case definition were included (sudden onset of symptoms and at least one of the following four systemic symptoms: fever or feverishness, malaise, headache, myalgia; and at least one of the following three respiratory symptoms: cough, sore throat, shortness of breath) [20].In all study sites, practitioners swabbed all elderly (60 or 65 years old and older) consulting for ILI, except for France where a proportion of elderly consulting for ARI were systematically selected for swabbing.Practitioners systematically selected patients from other age groups to swab using statistical sampling, except for Romania, where all patients consulting for ILI were swabbed.Hungary restricted their study population to those aged 18 years and over.
All participants in the study gave oral or written consent, in adherence with country requirements for ethical approval at each study site.The study period began 15 days after the start of the respective 2011/12 seasonal influenza vaccination campaign in each country.
Practitioners used standardised country-specific questionnaires to collect information on ILI signs and symptoms, sex, age, seasonal influenza vaccination in the 2011/12 and 2010/11 seasons, pregnancy, chronic conditions (including obesity, as defined in the participating countries), number of hospitalisations for chronic conditions in the past 12 months, receipt of antivirals (Spain and France excluded), and number of general practitioner (GP) visits in the past 12 months.Study sites included a question on belonging to the target group for vaccination, apart from France and Portugal, where this information was gathered using information on age, chronic conditions, and pregnancy.In addition, information related to target groups for vaccination was gathered in Portugal on whether the patient was a health professional or carer and a co-habitant or carer of a patient at-risk aged less than six months.
Among ILI patients fulfilling the inclusion criteria, we defined a case of influenza as a study participant whose swab tested positive for influenza virus by reverse-transcription polymerase chain reaction (RT-PCR) or culture.We classified patients with swabs testing negative for influenza virus as controls.
Swabs were tested for influenza at the respective country's National Influenza Reference Laboratory.In France, Italy, and Spain, tests were also conducted in other laboratories participating in the National Influenza Sentinel Surveillance System.At all study sites a subset of isolates were genetically and/or antigenically characterised.Details of laboratory viral detection, typing, subtyping and variant analysis performed are described elsewhere [21].
We defined a person as vaccinated if they had received at least one dose of 2011/12 seasonal influenza vaccine more than 14 days prior to ILI/ARI symptom onset.All the others were classified as unvaccinated.
The eight study teams sent their data to EpiConcept, where they were pooled and analysed.We carried out an analysis restricted to the A(H3) influenza type.We excluded controls presenting to the practitioner before the week of symptom onset of the first case and after the last case of influenza A(H3) in each country respectively.We restricted the study population to the target groups for vaccination.We compared the characteristics of cases and controls using chi-square tests, t-tests, Fisher's exact test or the Mann-Whitney test depending on the nature of the variable.
We used Cochran's Q-test and the I2 index to test the heterogeneity between study sites [22].
We estimated the pooled IVE as 1 minus the odds ratio (OR) of being vaccinated in cases versus controls, using a one-stage method with study site as fixed effect in the model.

Results
In the eight participating countries, influenza peaked at different times -from week 5 in Italy and Poland to week 10 in Portugal (Figure 1).
A total of 16 vaccines were used at the country study sites, of which four were adjuvanted.The median age was higher among cases (62.0 years; interquartile range (IQR): 37-70 years) than among controls (58.0 years; IQR: 41-69 years) (Table 2).
The proportion of cases presenting with any of the following symptoms was higher than controls: fever, malaise, myalgia and cough.A greater proportion of controls than cases had heart disease or at least one chronic condition.A greater proportion of controls visited their practitioner five or more times in the previous 12 months.A greater proportion of cases were swabbed within three days of symptom onset, but this was not statistically significant to the 5% level.The delay between vaccination and symptom onset was shorter for controls (median: 88.The Q test (p=0.142)and the I2 index (37.6%)testing for heterogeneity between the individual crude IVE estimates of the seven study sites included, suggested low to medium statistical heterogeneity.
In the early phase of the season (week 46/2011 to week 6/2012) the adjusted IVE was 38.1% (95% CI: -7.9 to 64.5) and in the late phase -0.7% (95% CI: -59.8 to 36.5).The adjusted IVE among persons with onset of symptoms less than three months since vaccination was 46.8% (95% CI: 9.0 to 68.9) and the IVE among persons with onset of symptoms three months or more since vaccination was 10.5% (95% CI: -32.5 to 39.5) (Figure 3).
When restricting to the early influenza phase, IVE among persons with onset of symptoms less than 93 days since vaccination was 48.4% (95% CI: -3.5 to 74.3) and the IVE among persons with onset of symptoms 2) and the IVE among persons with onset of symptoms more than 93 days since vaccination was 1.0% (95% CI: -60.5 to 38.9) (Figure 3).

Discussion
The overall adjusted pooled IVE estimates against influenza A(H3) from the multicentre case-control study in Europe among those targeted for vaccination was 24.8%, ranging between 15.1% in the elderly and 63.3% in persons aged between 15 and 59 years.This suggests a low adjusted IVE against medically attended A(H3) influenza among the target population except among younger adults.
The A(H3) strain was also predominant during the 2008/09 season, the I-MOVE pilot season.In that season, persons aged 65 and above had an IVE of 56.4% (95% CI: -0.2 to 81.0) against A(H3) [5].We observed a lower IVE in the 2011/12 A(H3) dominated season with an IVE of 15.1% in those aged 60 years and above and an IVE of 12.4% in those aged 65 years and above.
The strength of this study lies in its multicentre nature, enabling recruitment of a large sample size of participants across the EU.It is possible to restrict to the target group for vaccination and to stratify further by influenza type and age.Study sites adhere to a common protocol and carry out systematic sampling.They also collect information on potentially important positive and negative confounders.In addition, data quality is very high with only 1.7% (n=17/1033) of records with missing data.
Due to the observational nature of this study, we cannot exclude biases.We used a test-negative design, which is subject to the usual selection biases particularly for the control group.Study participants are selected according to a systematic sampling procedure by practitioners, who are blinded to the case and control status of the patients.This should minimise selection bias.
As I-MOVE is based on existing sentinel networks, GPs recruited patients according to the case definitions used in their network: the EU ILI case definition or the ARI case definition.As the ARI case definition is a more sensitive case definition than the EU ILI one, we could restrict the analysis to patients meeting the EU ILI case definition for all patients included in the study.
The test-negative design is a commonly used, but not validated study design [24][25][26][27][28][29][30][31][32].Using test-negative controls is considered to adjust for healthcare-seeking behaviour more so than if community controls were selected, as vaccination coverage varies by healthcareseeking behaviour [34,35].In addition, the covariate 'number of GP visits in the past 12 months' may adjust further for healthcare-seeking behaviour.Despite this adjustment, it is still debatable if test-negative controls properly reflect the vaccine coverage of the source population for cases [33].
While a higher proportion of controls visited their GP more frequently and had a chronic condition than cases, these variables were not strong confounders (-2% and 1% relative difference of IVE between model containing and not containing these confounders respectively).The main confounder was age group, changing the IVE of the adjusted model by 11%.
We cannot exclude residual confounding, either by unmeasured confounders or by use of broad categories within given confounders.However, we used 10-year age bands to reduce residual confounding by age.While we used month of symptom onset as a covariate, the IVE differs only little if using week of symptom onset (24.8% compared to 23.4% for overall IVE).
We included patients who were swabbed within seven days of symptom onset and we observed that a higher proportion of controls were swabbed more than three days after symptom onset than cases, although the difference is not statistically significant.The probability of influenza detection decreases with time between onset and swabbing, although the rate of decrease may vary by patient characteristics [35][36][37][38].It is possible that some misclassification bias is introduced by including false negative controls through including patients with a greater delay between onset of symptoms and swabbing.However the difference is small if we compare our results to an analysis restricting the study population to persons swabbed three days or fewer since symptom onset (24.8% compared to 22.8% for overall IVE).
Our study is limited by a small sample size for the stratified analyses.Therefore precise estimates were not always possible, particularly among the youngest age group, who are often the least numerous target group for vaccination.Estimates by influenza phase and by time since vaccination are also limited by the small sample size and although point estimates differ, confidence intervals overlap.
The majority of countries participating in this study used both adjuvanted and non-adjuvanted influenza vaccines.The different vaccine types were used in different subpopulations.With the data collected for this study, it was not possible to identify the target groups to enable an estimate by vaccine type.IVE estimates arising from the total population were lower than the estimates from the target group for vaccination, e.g.overall adjusted IVE of 10.9% (95% CI: -16.2 to 31.7) among the total population (data not shown), compared to 24.8% (95% CI: -5.6 to 46.5) among the target group for vaccination.We believe that the target group for vaccination is a more homogeneous study population in relation to vaccination, the main exposure of interest, as study participants belonging to the target group for vaccination are likely to have a more equal access to vaccination than the total population.One limitation of restricting to this population is that it is identified through the practitioner questionnaires, which did not collect information on target group homogeneously across study sites.In particular information on healthy persons with professions targeted for vaccination may have been omitted from some countries.Despite these limitations, we believe that our study suggests a low adjusted IVE against medically attended A(H3) influenza among the target population except for among young adults in the 2011/12 influenza season.
The lower IVE observed this season compared to the previous A(H3) dominated season (2008/09) may be due to changes in circulating viruses and hence suboptimal antigenic match between the 2011/12 vaccine and circulating strains.WHO and the Community Network of Reference Laboratories (CNRL) report northern hemisphere circulating A(H3N2) viruses being genetically and antigenically distinguishable from the A/ Perth/16/2009 vaccine strain and being more related to A/Victoria/361/2011-like reference viruses, differences which may have increased along the season [18,39].This virological change could have contributed to the lower IVE in the latter part of the season.
As the 2011/12 influenza season was a late season, persons presenting with influenza had a long delay between onset of symptoms and the vaccination, as campaigns were carried out in the autumn of 2011.The observed fall in IVE may also be due in part to waning of the immunity induced by the vaccine, perhaps markedly so in older people [40][41][42][43].Persons vaccinated less than 93 days before symptom onset showed a higher IVE than persons vaccinated 93 days or more before symptom onset.However, persons vaccinated 93 days or more before symptom onset were more likely to present later in the season, co-temporal with the emergence of antigenically drifted influenza viruses.To disentangle the possible effects of waning immunity and antigenic drift, we looked at IVE by early and late influenza phase.In the early influenza phase IVE was higher among persons vaccinated less than 93 days before symptom onset compared to persons vaccinated 93 days or more before symptom onset.This was not the case in the late influenza phase, where we may expect a greater effect of antigenic drift on the IVE estimates.This suggests the waning immunity hypothesis may be plausible.
In conclusion, the I-MOVE multicentre case-control study suggests a low IVE against medically attended A(H3) influenza in the 2011/12 season.The I-MOVE multicentre case control study provides high quality and rapid IVE estimates and should supplement the virological information that informs the WHO recommendations on vaccine strain selection [6,8].It is difficult to disentangle the respective roles of changes in the circulating viruses, possible waning immunity and otherwise imperfect vaccine.Further virological studies are needed on an annual basis quantifying drift over time as well as large epidemiological studies by time since vaccination with several delay categories to fully understand these potentially important issues.Production of an improved seasonal influenza vaccine with greater effectiveness should be given a high priority.

Figure 1
Figure 1Laboratory detection of influenza by week and virus subtype, Canada, 2012/13 sentinel surveillance system (n=833)

Table 1
Details for influenza A and B cases and controls originally considered for the mid 2012/13 season trivalent seasonal influenza vaccine effectiveness analysis, United Kingdom, 1 October 2012-4 January 2013 (n=1,865)

Table 2
Selecting participants with known symptom onset of influenza-like illness (n=1,865) for the 2012/13 trivalent seasonal vaccine effectiveness analysis, United Kingdom, 1 October 2012-4 January 2013

Table 3
Number of cases versus controls for influenza A and/or B according to 2012/13 trivalent seasonal influenza vaccine vaccination status and vaccine effectiveness (crude and adjusted a ) estimates, United Kingdom, 1 October 2012-4 January 2013 a Adjusted for age-group, sex, month and surveillance scheme.

Table 2
Profile of participants included in primary analysis, interim 2012/13 influenza vaccine effectiveness evaluation, Canada

Table 3
Laboratory profile of specimens included in primary analysis, interim 2012/13 influenza vaccine effectiveness evaluation, Canada

Table 5
Changes in amino acid sequence encoded by haemagglutinin (HA1) gene (antigenic regions) a for subset of 2012/13 Canadian sentinel influenza A(H3N2) strains relative to reference strains b Amino acid number HA1 45 48 53 54 62 67 88 94 121 124 128 142 145 156 157 186 192 198 219 230 278 280 304 312 Finally, in reviewing participant profiles, we identified no obvious signals of bias and in our analysis we adjusted for recognised potential confounders, but ultimately, given the observational design, we cannot rule out other unrecognised influences on the VE estimates.TLK and SS provided by a grant of the Canadian Institutes of Health Research.No other authors have competing interests to declare.
Although we have assessed vaccine relatedness through gene sequencing of communitybased sentinel viruses available from each province and across the season to date, in this interim assessment the sampling frame for specimen selection was not random or systematic.Bias may result from the preferential inclusion of specimens that demonstrate low cycle threshold values (high RNA levels) or successful virus isolation.These, however, are issues for all laboratory-based influenza surveillance. of 97% of the 642,051 inhabitants of the region (private companies provide healthcare to the remaining 3% of the population).The clinical records have been computerised since the year 2000 and include reports from primary care, hospital admissions, vaccination register, and laboratory test results.
present study was based on electronic clinical records in the region of Navarre, Spain in the 2011/12 season.The Navarre Ethical Committee for Medical Research approved the study protocol.The Navarre Health Service provides healthcare, free at point of service, to carried out a case-control study nested in the cohort of the population covered by the Navarre Health Service.Healthcare workers, persons living in nursing homes and children under six months of age were excluded.The study began in the first week in which influenza virus was detected, 12 December 2011 (week 50), and ended in the last week in which ILI patients tested positive for influenza, 20 May 2012 (week 20).

Table 2
Influenza vaccine effectiveness in preventing laboratory-confirmed influenza by patient characteristic, comparisons of influenza-positive cases (n=411) and test-negative controls (n=346), Navarre, 12 Dec 2011-20 May 2012 a Vaccine effectiveness adjusted for sex, age (<5; 5-14; 15-44; 45-64; ≥65 years), major chronic conditions, outpatient visits in the previous year, hospitalisation in the previous year, healthcare setting, and period of diagnosis.b There was one case of influenza A(H1N1)pdm09, not shown in this table.c Target population for vaccination includes people ≥60 years-old and people with major chronic conditions.

Table 3
Influenza vaccine effectiveness in preventing laboratory-confirmed influenza by vaccination status and time after vaccination, comparison of influenza-positive cases (n=411) and test-negative controls (n=346), Navarre, 12 Dec 2011-20 May 2012 a Vaccine effectiveness adjusted for sex, age (<5; 5-14; 15-44; 45-64; ≥65 years), major chronic conditions, outpatient visits in the previous year, hospitalisation in the previous year, healthcare setting, and period of diagnosis.b Target population for vaccination includes people ≥60 years-old and people with major chronic conditions.

Table 4
Distribution of influenza cases by type of virus and distribution of cases with characterisation by strains in two calendar periods.Navarre, Spain, 2011-2012

Table 2
Details for influenza A(H3N2) and B cases and controls, United Kingdom, October 2011-April 2012 (n=3,869) a Where a date of vaccination was missing this was estimated by assuming vaccination was on 19 October 2011, the median time of vaccination in controls with onset in 2012.

Table 5
Adjusted vaccine effectiveness estimates for influenza A(H3N2) by age, surveillance scheme and by time since vaccination, United Kingdom, October 2011-April 2012 (n=3,478) a Adjusted for age group, sex, month and surveillance scheme.
Influenza-like illness / acute respiratory infection rates by week of symptom onset as reported by the national sentinel systems, I-MOVE multicentre case-control study, study sites in eight European Union countries, influenza season 2011/12 ARI: acute respiratory infection; ILI: influenza-like illness.

Table 1
Participating practitioners and recruited influenza-like illness patients, by A(H3) influenza case-control status, vaccination status and study site, multicentre case-control study, study sites in eight European Union countries a , 2011/12 ILI: Influenza-like illness; ISO: International Organization for Standardization.NA: not applicable a France, Hungary, Ireland, Italy, Poland, Portugal, Romania, Spain b ILI patients meeting the European Union case definition, swabbed less than eight days after onset of symptoms.c From 15 days after the start of the vaccination campaign; controls with an onset of symptoms in the weeks prior to the first influenza A(H3) case or after the last influenza A(H3) case were excluded.d ILI patients in a target group for vaccination included in the study, after excluding those with missing information on laboratory results, vaccination status or date of vaccination.
ILI: influenza-like illness.aFrance, Hungary, Ireland, Italy, Portugal, Romania, Spain b Unless otherwise indicated.cNon parametric test of the median.dTwo-sided Fisher's exact test.eVaccination more than 14 days before onset of influenza-like illness symptoms.fAs defined in the respective countries.93

Table 3
Pooled crude and adjusted 2011/12 seasonal influenza vaccine effectiveness against laboratory-confirmed A(H3) influenza in vaccination target groups, at study sites in seven European Union countries a , week 46/2011-week 17/2012 (patients with complete information, n=1,016) a France, Hungary, Ireland, Italy, Portugal, Romania, Spain b Study site included in the model as fixed effect.c Model adjusted for presence of at least one chronic disease, sex, at least one hospitalisation for chronic disease in the previous 12 months, age group, practitioners' visits in the previous 12 months (0-1, 2-4 and ≥5 visits), month of symptom onset and study site.d November dropped due to no cases (two records dropped).