2015/16 I‐MOVE/I‐MOVE+ multicentre case‐control study in Europe: Moderate vaccine effectiveness estimates against influenza A(H1N1)pdm09 and low estimates against lineage‐mismatched influenza B among children

Background During the 2015/16 influenza season in Europe, the cocirculating influenza viruses were A(H1N1)pdm09 and B/Victoria, which was antigenically distinct from the B/Yamagata component in the trivalent influenza vaccine. Methods We used the test‐negative design in a multicentre case‐control study in twelve European countries to measure 2015/16 influenza vaccine effectiveness (VE) against medically attended influenza‐like illness (ILI) laboratory‐confirmed as influenza. General practitioners swabbed a systematic sample of consulting ILI patients and a random sample of influenza‐positive swabs was sequenced. We calculated adjusted VE against influenza A(H1N1)pdm09, A(H1N1)pdm09 genetic group 6B.1 and influenza B overall and by age group. Results We included 11 430 ILI patients, of which 2272 were influenza A(H1N1)pdm09 and 2901 were influenza B cases. Overall VE against influenza A(H1N1)pdm09 was 32.9% (95% CI: 15.5‐46.7). Among those aged 0‐14, 15‐64 and ≥65 years, VE against A(H1N1)pdm09 was 31.9% (95% CI: −32.3 to 65.0), 41.4% (95% CI: 20.5‐56.7) and 13.2% (95% CI: −38.0 to 45.3), respectively. Overall VE against influenza A(H1N1)pdm09 genetic group 6B.1 was 32.8% (95% CI: −4.1 to 56.7). Among those aged 0‐14, 15‐64 and ≥65 years, VE against influenza B was −47.6% (95% CI: −124.9 to 3.1), 27.3% (95% CI: −4.6 to 49.4) and 9.3% (95% CI: −44.1 to 42.9), respectively. Conclusions Vaccine effectiveness (VE) against influenza A(H1N1)pdm09 and its genetic group 6B.1 was moderate in children and adults, and low among individuals ≥65 years. Vaccine effectiveness (VE) against influenza B was low and heterogeneous among age groups. More information on effects of previous vaccination and previous infection is needed to understand the VE results against influenza B in the context of a mismatched vaccine.


| INTRODUCTION
Nine of twelve study sites also participated in a pilot laboratory project where they randomly selected specimens for sequencing of at least the gene segment coding for the haemagglutinin, in order to compute a representative VE estimate against the influenza A(H1N1) pdm09 6B.1 genetic group.

| METHODS
Twelve European study sites located in Croatia, France, Germany, Hungary, Ireland, Italy, Poland, Portugal, Romania, Spain, Sweden and the Netherlands participated in the test-negative 2015/16 multicentre case-control study. The methods have been described previously [2][3][4] Background: During the 2015/16 influenza season in Europe, the cocirculating influenza viruses were A(H1N1)pdm09 and B/Victoria, which was antigenically distinct from the B/Yamagata component in the trivalent influenza vaccine.

Methods:
We used the test-negative design in a multicentre case-control study in twelve European countries to measure 2015/16 influenza vaccine effectiveness (VE) against medically attended influenza-like illness (ILI) laboratory-confirmed as influenza. General practitioners swabbed a systematic sample of consulting ILI patients and a random sample of influenza-positive swabs was sequenced. We calculated adjusted VE against influenza A(H1N1)pdm09, A(H1N1)pdm09 genetic group 6B.1 and influenza B overall and by age group.  and are based on the ECDC generic case-control study protocol and the I-MOVE+ protocol. 5,6 Participating practitioners interviewed and collected nasopharyngeal or combined naso-and oropharyngeal specimens from a systematic sample of consenting patients seeking medical attention for influenza-like illness (ILI). In Hungary, only patients aged 18 years and older and in Croatia only patients aged 65 years and older were eligible. Practitioners collected in a standardised report form information including symptoms, date of onset and swabbing, 2015/16 seasonal vaccination status, date of influenza vaccination and vaccine product, prior (2014/15) seasonal vaccination status, sex, age and presence of chronic medical conditions in the past 12 months.

Results
Seven study sites included a question on belonging to the target group for vaccination. In France, Germany, Poland, Portugal and Sweden, the target group was defined from patients' information on age, chronic conditions and pregnancy. Additionally, in Portugal, being a health professional or carer and a cohabitant or carer of a patient at risk aged less than 6 months and in Poland, belonging to an occupational risk group (eg, healthcare worker), defined the target group.
In the pooled analysis, we included patients meeting the European Union ILI case definition, 7 swabbed within 7 days of symptom onset, and who had not received antivirals in the 14 days prior to swabbing.
A case of confirmed influenza was an ILI patient who was swabbed and tested positive for influenza virus using real-time reverse-transcription polymerase chain reaction (RT-PCR). Controls were ILI patients who tested negative for any influenza virus using RT-PCR.
We defined a person as vaccinated if he or she had received at least one dose of a 2015/16 seasonal influenza vaccine more than 14 days before ILI symptom onset. Those vaccinated less than 15 days before ILI onset were excluded. All other patients were classified as unvaccinated.
For each study site, we included ILI patients presenting more than 14 days after the start of national or regional influenza vaccination campaigns and we excluded controls presenting before the onset week of the first influenza type/subtype-specific case. ILI patients presenting in weeks of onset after two or more consecutive weeks of no cases and influenza A cases that were not further subtyped were also excluded from the analysis.
For each study site, we computed the odds ratio (OR) of being vaccinated in cases vs controls. We conducted a complete analysis excluding patients with missing values for any of the variables in the model measuring adjusted VE. Using Cochran's Q-test and the I 2 index, we tested the heterogeneity between study sites. 8 We estimated the pooled type/subtype influenza VE as (1 minus the OR)*100 using a one-stage model with study site as a fixed effect.
Using a logistic regression model, we calculated VE including potential confounding factors: date of symptom onset (modelled as a restricted cubic spline with 4 knots where sample size allowed), age (modelled as a restricted cubic spline with 4 knots or age groups depending on the analysis), sex and presence of at least one chronic medical condition (including pregnancy and obesity where available).
We used the one in ten rule of predictor degrees of freedom to events to determine the maximum number of covariates to include in analyses with low sample sizes in order to avoid overfitting the model. 9,10 To study the effect of prior ( To study the effects of waning on the vaccine effect within a season, we further estimated VE by time since vaccination, modelling days between vaccination and symptom onset dates as a restricted cubic spline with 4 knots. 11 In this analysis, we additionally included patients vaccinated 14 days or less before symptom onset (excluded from the main analysis).
Nine study sites participated in a laboratory pilot project (DE, FR, HU, IE, PT, RO, SE, ES and NL) for sequencing at least the haemagglutinin gene segment for each influenza type/subtype. In this laboratory pilot project, either all specimens were selected for sequencing or a proportion of specimens were randomly selected for sequencing to ensure representativity. The proportion of specimens randomly selected for sequencing could vary over time (eg, higher early in the season and lower during the peak) and a sampling fraction was calculated for each study site and time unit. The specimens were sent to the corresponding National Influenza Centre, where influenza diagnosis was confirmed, and viruses were characterised by sequencing the HA1 coding portion of the haemagglutinin gene. Analysis of the nucleotide and amino acid sequences of the HA1 coding portion of the haemagglutinin gene was performed in MEGA6 to determine clade distribution.
We weighted the genetic group-specific VE analysis using the reciprocal of the sequencing sampling fraction for each time period and study site and used robust standard errors.
Data management and statistical analyses were carried out using

| RESULTS
The 2015/16 influenza season in Europe was characterised by the The proportion vaccinated with the 2015/16 influenza vaccine was 9.7% among controls, 6.7% among influenza A(H1N1)pdm09 cases and 6.3% among influenza B cases ( Table 1).
The median age of influenza A(H1N1)pdm09 cases was 35 years, of controls 29 years and of influenza B cases 12 years (Table 2).
Compared to influenza A(H1N1)pdm09, a higher proportion of influenza B cases were less than 15 years (55.3% vs 30.3%) and a lower proportion were 15-64 years old (40.8% vs 63.5%). The proportion of patients aged 65 and older varied between controls, influenza A(H1N1) pdm09 and influenza B cases with 9.5%, 6.2% and 3.9%, respectively.
The proportion of patients with at least one chronic condition was similar between controls and influenza A(H1N1)pdm09 cases (20.2% and 17.6%, respectively), but lower among influenza B cases (11.9%).
Among controls, 81.7% were swabbed within 3 days of symptom onset compared to 84.9% and 85.2% of influenza A(H1N1)pdm09 and influenza B cases, respectively. Among controls, 6.5% were swabbed on the day of symptom onset, compared to 4.2% and 4.3% of influenza A(H1N1)pdm09 and influenza B cases, respectively.

| Influenza A(H1N1)pdm09
Statistical heterogeneity between VE estimates against influenza A(H1N1)pdm09 by study site was low overall (among all ages) and among those aged 15-64 years (I 2 index 0% and 10%, respectively).
Due to small sample sizes, it was not possible to estimate heterogeneity among other age groups.

| Influenza B
The I 2 index for heterogeneity between VE estimates against influ-   f Due to heterogeneity of VE estimates against influenza B between age groups, no "all ages" estimate against influenza B was attempted.

Number of influenza B cases/total n (%)
g Adjusted by time and study site only.
T A B L E 2 (Continued) injectable vaccine (−1%). 22 In the USA, there is a universal vaccination recommendation, and in the UK and Finland, vaccine is recommended in certain age groups in children. However, in the countries participating in the I-MOVE/I-MOVE+ multicentre case-control study, vaccine is recommended only to children with chronic conditions, with the exception of Poland where vaccination is recommended among those aged 6 months to 18 years. [23][24][25] The The VE point estimate was higher for subunit vaccine than for split virion vaccine among children, but precision was low. Both estimates were low, indicating that the low VE was not due to a vaccine typespecific issue. Among 15-to 64-year-olds, split virion VE point estimate was higher than subunit vaccine, but again precision was low.