Effectiveness of trivalent seasonal influenza vaccine in preventing laboratory-confirmed influenza in primary care in the United Kingdom : 2012 / 13 end of season results

N Andrews (Nick.Andrews@phe.gov.uk)1, J McMenamin2, H Durnall3, J Ellis4, A Lackenby4, C Robertson2,5,6, B von Wissmann2, S Cottrell7, B Smyth8, C Moore7, R Gunson9, M Zambon4, D Fleming3, R Pebody1 1. Public Health England Health Protection Directorate, Colindale, London, United Kingdom 2. Health Protection Scotland, Glasgow, United Kingdom 3. Royal College of General Practitioners Research and Surveillance Centre, Birmingham, United Kingdom 4. Public Health England Operations Directorate, Microbiology Services, Colindale, London, United Kingdom 5. University of Strathclyde, Glasgow, United Kingdom 6. International Prevention Research Institute, Lyon, France 7. Public Health Wales, Cardiff, United Kingdom 8. Public Health Agency Northern Ireland, Belfast, United Kingdom 9. West of Scotland Specialist Virology Centre, Glasgow, United Kingdom


Introduction
The 2012/13 influenza season in the United Kingdom (UK) was unusually long with elevated levels of activity persisting from week 50 to 16.In England, Northern Ireland and Wales, the season was dominated initially by circulation of influenza B virus, with school outbreaks in the period before Christmas.This was followed by influenza A(H3N2) virus circulation particularly in the New Year and spring with influenza outbreaks in often highly vaccinated care home populations [1].Scotland presented a different picture with influenza activity initially dominated by influenza A(H3N2) followed by influenza B virus circulation.
The occurrence of late season influenza A outbreaks in much of the UK some months after the completion of the 2012/13 influenza vaccine campaign led to questions being raised about waning intra-seasonal vaccine protection.A similar observation of late season influenza A(H3N2) outbreaks in care home settings in 2011/12 was accompanied by the observation of significant intra-seasonal waning in protection for those vaccinated more than three months previously [2].Trivalent seasonal influenza vaccine (TIV) in 2012/13 included an A/California/7/2009 (H1N1)pdm09-like virus, an A/Victoria/361/2011 (H3N2)-like virus and a B/Wisconsin/1/2010-like virus (Yamagata lineage).In 2012/13, vaccine uptake was 73.4% in those aged over 65 years and 51.3% in risk groups with individuals aged under 65 years in England [1].
Mid-season estimates from the UK and elsewhere were published in January 2013 with vaccine effectiveness (VE) against all influenza types ranging from 45% in Canada to 50% in a European study and 51% in the UK [3][4][5].This study presents the end-of-season VE for the 2012/13 seasonal TIV in preventing medically attended confirmed influenza A(H3N2), A(H1N1)pdm09 and B virus infection using the established primary care sentinel swabbing surveillance schemes in the UK [2,3,6].It also examines the protective effect of vaccination measured at different points during the season and by time since vaccination, to determine if there is any evidence of intra-seasonal waning protection.

Study population and period
Data were derived from five primary care influenza sentinel swabbing surveillance schemes in the UK from England (two schemes), Northern Ireland, Scotland, and Wales.Details of the Royal College of General Practitioners (RCGP), Public Health England (PHE) Specialist Microbiology Network (SMN), Public Health Agency (PHA) of Northern Ireland, Health Protection Scotland (HPS) and Public Health Wales swabbing schemes have been published previously [7].
The study period ran from 1 October 2012 to 24 April 2013.Patients were swabbed as part of clinical care, with verbal consent.Cases were defined, 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 B. ILI was defined as an individual presenting in primary care with an acute respiratory illness with physician-diagnosed fever or complaint of feverishness.Controls were individuals presenting with ILI in the same period who were swabbed and tested negative for influenza.
A standardised questionnaire was completed by the GP responsible for the patient during the consultation.Demographic, clinical and epidemiological information was collected from cases and controls, including date of birth, sex, defined underlying clinical risk group, date of onset of respiratory illness, date of specimen collection, and influenza vaccination status for the 2012/13 season with vaccination dates.

Laboratory methods
Laboratory confirmation was undertaken using realtime polymerase chain reaction (RT-PCR) assays capable of detecting circulating influenza A viruses, influenza B viruses and other respiratory viruses [8,9].Samples in England were sent to the PHE Microbiology Services, Colindale (RCGP scheme) or one of the specialist PHE microbiology laboratories (SMN scheme).Samples in Northern Ireland were sent to the Regional Virus Laboratory, Belfast, in Scotland to the West of Scotland Specialist Virology Centre, Glasgow (HPS scheme), and in Wales to the Public Health Wales Specialist Virology Centre, Cardiff.
Further strain characterisation was also performed.Influenza viruses were isolated in Madin-Darby canine kidney (MDCK) or MDCK-SIAT1 cells from all RT-PCR positive samples from England as previously described [10].Virus isolates with a haemagglutination titre ≥40 were then characterised antigenically using post-infection ferret antisera in haemagglutination inhibition (HI) assays, with guinea pig (A(H3N2) viruses) or turkey (A(H1N1)pdm09 and influenza B viruses) red blood cells [11].Nucleotide sequencing of the HA1 region of the haemagglutinin (HA) gene of a subset of influenza B viruses was performed (primer sequences available on request), and phylogenetic trees were constructed with a neighbour-joining algorithm available in the  1).

Statistical methods
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 from analysis.If the date of vaccination was missing, as the 2012/13 campaign occurred before influenza circulation was common, it was assumed that TIV vaccination was more than 14 days before onset date.Those with a missing date of onset or an onset date more than seven days before the swab was taken were excluded.
VE was estimated by the test-negative case-control (TNCC) design [12].In this design VE is calculated as 1 -(odds ratio) obtained using multivariable logistic regression models with influenza A or influenza B PCR results as outcomes and seasonal vaccination status as the linear predictor.Influenza A results were also further categorised for A(H3N2) and A(H1N1)pdm09 viruses.In the analyses evaluating VE for a specific type or strain, those positive for other types were excluded.Age (coded into five standard age groups, <5, 5-

Strain characterisation
Antigenic analysis by HI assay of influenza A(H3N2) viruses isolated from positive samples submitted through the RCGP scheme, demonstrated that these viruses were antigenically homogeneous, and closely related antigenically (<fourfold differences in HI assays for all 89 tested) to the cell-propagated H3N2 vaccine strain, A/Victoria/361/2011 [1].The relatively fewer influenza A(H1N1)pdm09 viruses isolated in 2012/13 were closely related antigenically to the A(H1N1) pdm09 2012/13 vaccine strain, A/California/7/2009, though six of thirty isolates did show reduced reactivity in antigenic characterisation assays with antiserum raised against influenza A/California/7/2009 (>fourfold difference in HI assays).

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, except sex, were significantly associated with having a positive swab, only age group was a confounder for the vaccine effects (changed the estimate by more than 5%).Information on risk group was missing for 182 of 3,286 samples (5.5%) and was therefore not included in the final model.If risk group was included, it was found not to be associated with being positive and the VE estimates remained similar.
Tables 3 and 4 show VE estimates against influenza A(H3N2), A(H1N1)pdm09 and B according to vaccination status and time since vaccination and period.
For influenza A(H1N1)pdm09, VE showed evidence of a decline by time since vaccination (Table 4).For influenza A(H3N2), there was also evidence of a decline in VE from 50% (95% CI: 16 to 71) within three months of vaccination to 2% (95% CI: -49 to 36) after three months.This decline was not, however statistically significant (p=0.25) and was less apparent when looking at VE by period (Table 3).Assessing time since vaccination with those with a missing date of vaccination excluded gave similar results with a decline from 100% (95%CI: 56 to 100) to 63% (95% CI: -7 to 87) for influenza A(H1N1) pdm09 and from 41% (95%CI:0 to 66) to 5% (95% CI: -51 to 41) for influenza A(H3N2).
The adjusted age-specific estimate for influenza A protection was lower in the oldest age group (≥65 years compared to other ages) (Table 4), however, the observed differences were not significant.VE in the vaccine target group (aged ≥65 years or in a risk group) was 13% (95% CI:-44 to 67) due to the low VE estimate in those aged ≥65 years.VE also showed some Differences between cases and controls for all variables in this table were statistically significant, chi-squared test.a Two individuals positive for influenza A(H3N2) and B and one individual positive for A(H1N1)pdm09 and B, are included in both the influenza B and influenza A columns.For the totals in this column, these individuals are only counted once.b 25 influenza A cases were of unknown strain, these are included in influenza A VE analysis but not the strain specific analyses.Also two individuals positive for influenza A(H3N2) and B and one individual positive for A(H1N1)pdm09 and B are included in both the influenza B and influenza A columns, which is why the total adds to 3,289 for controls, influenza A and influenza B. c Where a date of vaccination was missing (n=150) this was estimated by assuming vaccination was on 20 October 2012, the median time of vaccination in controls with onset in 2013.variability across the schemes although this difference was not significant.

Vaccine effectiveness against influenza B infection
The adjusted VE of TIV against influenza B was 51% (95% CI: 34 to 63).VE was 57% within three months of vaccination and non-significantly lower at 42% after three months (Table 4).VE did not vary by age group or scheme.

Discussion
In this study we found moderate effectiveness of 2012/13 TIV against laboratory-confirmed influenza B and good protection against influenza A(H1N1)pdm09 infection.However, VE against influenza A(H3N2) infection was poor at only 26% (95% CI: -4 to 48).We also found a non-significant trend that effectiveness waned by time since vaccination for influenza A(H3N2), which is consistent with the waning seen against influenza A(H3N2) for the 2011/12 TIV vaccine in the previous season where it reduced from 53% (95% CI: 0 to 78) within three months of vaccination to 12% (95% CI: -31 to 41) after three months [2].The point estimate for 2011/12 was also similar for influenza A(H3N2) at 23% (95% CI: -10 to 47), although VE was higher in 2011/12 against influenza B at 92% (95% CI: 38 to 99) compared to 2012/13.The VE against A(H1N1)pdm09 is consistent with that seen with monovalent pandemic vaccine (adjuvanted) in 2009/10 and with TIV in 2010/11 [6,7].
Influenza vaccine strains are propagated in eggs during the vaccine manufacturing process.It has been reported that propagation of the A/Victoria/361/2011(H3N2)-like vaccine viruses for vaccine production resulted in antigenic changes in the virus resulting from adaptation to the growth in eggs, although circulating viruses were closely related antigenically to the cell-propagated influenza A(H3N2) vaccine strain, A/Victoria/361/2011 [13].Following these observations the influenza A(H3N2) vaccine component for use in the 2013/14 season (northern hemisphere winter) has been updated to recommend an influenza A(H3N2) virus antigenically like the cell-propagated A/Victoria/361/2011 prototype strain (such as A/Texas/50/2012) [13].The majority of influenza B virus isolates in 2012/13 were characterised as belonging to the B/Yamagata/16/88-lineage, as does the 2012/13 influenza B vaccine strain, B/ Wisconsin/1/2010.However, genetically, the HA genes of the majority of B-Yamagata strains fell within a genetic clade (clade 2), which in HI assays are antigenically distinguishable from B/Wisconsin/1/2010 (genetic clade 3)-like viruses.This provides an explanation for the lower VE observed against influenza B in compared to 2011/12.Consequently the influenza B vaccine component recommended for use in 2013/14 has also been updated, to a B/Massachusetts/2/2012 (clade2)-like virus [13].
When stratifying VE by age and scheme, VE is estimated with lower precision.There were no significant differences in VE by age or scheme although the point estimate for VE against influenza A was negative for the SMN scheme and also for the over 65 years age group.These differences are likely to be chance fluctuations due to small numbers and emphasise the need for large numbers of swabs for precise estimates for such subgroup analyses.This is now the fourth season in which a pooled UK VE analysis has been performed using the TNCC design with mid-season estimates also produced for 2009/10, 2010/11 and 2012/13 [2,3,6,7,14].Results from the RCGP scheme have also been published for 2005/06, 2006/07 and 2007/08 [15].The results of each season have been consistent with those published from other countries and from pooled European analyses but often with greater precision in the UK due to the large numbers of swabs.The mid-season 2012/13 results gave VE against influenza A of 49% (95% CI: -2 to 75) and influenza B of 52% (95% CI: 23 to 70), which was similar to other early season results from Europe and Canada [4,5], and also similar to the end of season result of 51% (95% CI: 34 to 63) for influenza B, but higher than the end of season result of 35% (95% CI: 11 to 53) for influenza A. The TNCC design is now the most commonly used method for estimating the VE of influenza vaccines.This reflects the advantages of the method in terms of its simplicity and the fact that those that test negative form an excellent control group as they are well matched on propensity to consult a GP.Further discussion of the methodological issues have been published previously [12,16,17] and a recent paper has demonstrated the methods validity compared to placebo controlled clinical trial results [18].
The intra-seasonal waning of VE against influenza A(H3N2) seen in 2011/12 in the UK is supported by the estimates seen this year, albeit non-significant.Carehome outbreaks late in the season both in 2011/12 and 2012/13 also support this observation [1,19], as did similar findings in other countries in 2011/12 [20].With a new influenza A(H3N2) and B strain recommended for 2013/14 and with the introduction of a trivalent live attenuated intranasal vaccine for all children aged 2-3 years and up to 10 years of age in parts of the UK [21] monitoring of VE remains an essential part of influenza surveillance.
collection.Funding: The infrastructure and methodology to enable the work was developed as part of a European Centre for Disease Control and Prevention (ECDC) funded project "I-MOVE".The I-MOVE group continues to be co-ordinated by Epi-Concept (http://www.epiconcept.fr/).

Figure 3
Figure 3Phylogenetic analysis with sequences from reference viruses downloaded from NCBI GenBank and GISAID EpiFlu databases of influenza B sequences derived from patients in the United Kingdom, 2012/13 influenza season

Table 1
Reference influenza B haemagglutinin 1 sequences used in phylogenetic analysis on Sharing Avian Influenza Data (GISAID) (Table

Table 2
Details for influenza A and B cases and controls, United Kingdom, October 2012-April 2013 (n=3,286 individuals)

Table 4
Adjusted vaccine effectiveness estimates for influenza by age, surveillance scheme and by time since vaccination, United Kingdom, October 2012-April 2013 CI: confidence interval; HPS: Health Protection Scotland; PHA: Public Health Agency; RCGP: Royal College of General Practitioners' surveillance scheme; SMN: Public Health England Specialist Microbiology Network, VE: vaccine effectiveness.a Adjusted for age group, sex, month of sample collection and surveillance scheme.b Number of vaccinated cases/controls too low to give an estimate with meaningful precision (95% CI lower end <-200 and upper end >80).c Unadjusted Cornfield 95% CI.

Table 3
Adjusted vaccine effectiveness estimates based on samples positive (cases, N=1,330) and negative (controls, N=1,956) for influenza according to vaccination status, United Kingdom, October 2012-April 2013 Because two individuals positive for influenza A(H3N2) and B, and one individual positive for A(H1N1)pdm09 and B are included in both the influenza B and influenza A cases, summing up the cases presented in this column amounts to 1,333 instead of 1,330.b Adjusted for age group, sex, month of sample collection and surveillance scheme. a