Vaccine effectiveness against influenza hospitalisation in adults during the 2022/2023 mixed season of influenza A(H1N1)pdm09, A(H3N2) and B circulation, Europe: VEBIS SARI VE hospital network

Abstract We conducted a multicentre hospital‐based test‐negative case–control study to measure vaccine effectiveness (VE) against PCR‐confirmed influenza in adult patients with severe acute respiratory infection (SARI) during the 2022/2023 influenza season in Europe. Among 5547 SARI patients ≥18 years, 2963 (53%) were vaccinated against influenza. Overall VE against influenza A(H1N1)pdm09 was 11% (95% CI: −23–36); 20% (95% CI: −4–39) against A(H3N2) and 56% (95% CI: 22–75) against B. During the 2022/2023 season, while VE against hospitalisation with influenza B was >55%, it was ≤20% for influenza A subtypes. While influenza vaccination should be a priority for future seasons, improved vaccines against influenza are needed.


| BACKGROUND
In February 2022, the World Health Organization (WHO) recommendations for the northern hemisphere 2022-2023 influenza season for trivalent influenza vaccines were to include an A/Wisconsin/588/2019 (H1N1)pdm09-like virus, an A/Darwin/6/2021 (H3N2)-like virus and a B/Austria/1359417/2021 (B/Victoria lineage)-like virus.Quadrivalent vaccine was recommended to contain the above three viruses plus a B/Phuket/3073/2013 (B/Yamagata lineage)-like virus. 1 Egg-based vaccines should contain the same last two viruses, in addition to an A/Victoria/2570/2019 (H1N1)pdm09-like and an A/Darwin/9/2021 (H3N2)-like virus. 1 Within the European Union/European Economic Area, an early seasonal epidemic threshold (10% positive sentinel specimens) was crossed by week 45, peaking at week 51, 2022. 2 There was co-circulation of influenza A(H3N2), A(H1N1)pdm09 and B viruses, initially predominated by A subtypes, although patterns of dominance varied across countries. 3There were hospitalised severe acute respiratory infection (SARI) cases (occurring mainly in older adults) caused by infection with each of these influenza types; however, ultimately, over the 2022/2023 season, influenza A(H1N1)pdm09 predominated among SARI patients. 2 pooled data from nine study sites in eight European countries participating in the ECDC Vaccine Effectiveness, Burden and Impact Studies (VEBIS) SARI vaccine effectiveness (VE) hospital network 3 (Figure 1) with sufficient data to estimate influenza VE against circulating strains.We estimated VE against hospitalisation due to PCRconfirmed influenza among SARI patients swabbed between 1 October 2022 and 15 May 2023.

| STUDY DESIGN AND DESCRIPTIVE RESULTS
Participating sites in the network use a common generic protocol 4,5 and the test-negative case-control design. 6Hospital teams collect demographic, clinical and influenza vaccination information via questionnaire, electronic medical records, vaccine registries or patient interview.We defined a SARI patient as an individual hospitalised for ≥24 h with at least one systemic symptom or sign: fever or feverishness, malaise, headache or myalgia or deterioration of general condition (asthenia, loss of weight or anorexia, confusion or dizziness) and at least one respiratory symptom or sign (cough, sore throat or shortness of breath) at admission or within 48 h after admission.Patients were excluded if their symptoms started (or clearly worsened, if chronic) more than 7 days before swabbing.
Cases were SARI patients PCR-positive and controls PCRnegative for influenza within 48 h of admission.We estimated VE overall, by age group (0-17, 18-64, ≥65 years), by target group for influenza vaccination according to each country's specific recommendations and by presence of at least one versus no chronic condition.We included 5547 SARI patients aged ≥18 years (886 cases of any influenza, 4661 test-negative controls) from 38 European hospitals, in nine of 13 participating study sites (Figure 1) providing sufficient data for this period.
Fifty-one per cent of cases and 49% controls were female; 68% of cases and 72% of controls had at least one of the five commonly collected chronic conditions (diabetes, heart disease, lung disease, asthma, immunosuppression).Forty per cent of cases and 56% controls were vaccinated against influenza (Table 1).

| VACCINATION DEFINITIONS AND VACCINE EFFECTIVENESS
We defined the start of the 2022/2023 influenza season in each participating site as the week number in which the first PCR-confirmed case was reported to the study.We defined vaccination as receipt of influenza vaccine ≥14 days after the start of the current season's vaccination campaign in each participating country.Only vaccines received ≥14 days before onset were considered valid (those vaccinated 1-13 days before onset were excluded; those vaccinated on or after onset were considered unvaccinated).SARI patients received their influenza vaccination between weeks 36, 2022 and week 3, 2023 (cases) or week 5, 2023 (controls); Figure 2. Vaccination dates coincided with the early start of this 2022/2023 influenza season (Figure 2).The median time from vaccination to onset was 74 days for cases, 88 days for controls.Where influenza vaccine type was known (n = 2017/2963; 68%), all SARI patients were vaccinated with quadrivalent vaccine (Table 1).
We compared the odds of vaccination between cases and controls using logistic regression, adjusting the odds ratio (OR) of vaccination among cases and controls by age, sex and presence of at least one of the five chronic conditions listed above.We calculated VE as 1 minus the adjusted OR (expressed as a percentage).We included study site (as a fixed effect) and date of onset (as a restricted cubic spline or categorical variable using onset month) in all analyses.We excluded sites with <10 cases or controls for each influenza (sub) type-specific analysis.We estimated interaction terms between vaccination and (1) age group and (2) chronic condition status.
Where the number of cases or controls per parameter was <10, we conducted sensitivity analysis using Firth's method of penalised logistic regression (PLR) to assess small sample bias. 7,8We considered estimates having an absolute difference >10% between the PLR and original VE estimate indicative of small sample bias and do not show these estimates.Similarly, we do not show estimates for analyses with <20 vaccinated cases or controls.As we expected most SARI patients to have been systematically tested for both influenza and SARS-CoV-2, we performed sensitivity analysis excluding all controls infected with SARS-CoV-2.
Vaccinated case numbers were too small for valid estimates by age group or chronic condition status against influenza B (Table 2).
The difference between VE estimates from the main and the sensitivity analysis excluding controls with SARS-CoV-2 infection was ≤9%, depending on influenza (sub)type (Table 2).

| DISCUSSION AND CONCLUSION
Our results from this multi-country European hospital study suggest that, for the 2022/2023 influenza season, adult SARI patients had the highest all-age VE against influenza B (56%).The VE was lower against influenza A(H3N2) (20%) and against influenza A(H1N1)pdm09 (11%).Point estimates were highest in older adults (≥65 years; 29% against influenza A) and those in the vaccine target group (66% against influenza B).
Despite the early season start, the vaccine coverage among controls (66% in the age group ≥65 years; data not shown) was similar to previous seasons. 9Our point estimates for VE against any influenza A and against A(H3N2) were similar (20%) to those found in our interim study (24%) 3 and to our 2017/2018 end-of-season VE estimates for influenza A(H3N2) in a hospital setting (24%). 9Results from a testnegative design case-control study in the United States and one in Italy for the 2022/2023 season reported higher overall VE results against influenza A and against A(H3N2) than ours, at 35% (95% CI: 27%-43%) 10 and 38% (95% CI: À34-74), 11 respectively.However, the VE in those aged ≥65 years in the Italian study was 24% (95% CI: À86-72), very close to the 25% (95% CI: À1-44) observed in our study.Although the US study included only SARS-CoV-2-negative controls, their VE for influenza A was still higher than those in our sensitivity analysis including only SARS-CoV-2-negative controls, although 95% CIs overlapped (26%; 95% CI: 10-39). 10In Europe, most circulating influenza A(H3N2) viruses belonged to haemagglutinin clade 2.2. 12   In analyses stratified by chronic condition, the adjustment for presence/absence of chronic condition was removed.d Seven sites: Belgium, Croatia, Lithuania, Malta, Navarra, Romania and Spain.e Fewer than 10 cases per parameter in the model.
f Five sites: Belgium, Germany, Malta, Navarra and Spain.g Four sites: Belgium, Lithuania, Navarra and Spain.h Six sites: Belgium, Croatia, Lithuania, Navarra, Romania and Spain.i Three sites: Belgium, Navarra and Spain.
Against influenza B, VE was 56% (95% CI: 22-75) overall.Against F I G U R E 1 Countries and study sites 1 participating in the Vaccine Effectiveness, Burden and Impact Studies (VEBIS) hospital network by provision of data for this analysis, Europe, 2022-2023 influenza season. 1 Participating sites: Belgium (BE), Croatia (HR), Czechia (CZ), France (FR), Germany (DE), Hungary (HU), Ireland (IE), Lithuania (LT), Malta (MT), Portugal (PT), Romania (RO), Spain (11 regions: ES; Navarra region: NA).Included in this analysis: BE, ES, DE, HR, LT, MT, NA, PT, RO.T A B L E 1 Patient characteristics of cases and controls, Vaccine Effectiveness, Burden and Impact Studies (VEBIS) hospital network, 2022-2023 influenza season, Europe (N = 5547). c Vaccine effectiveness against influenza hospitalisation, Vaccine Effectiveness, Burden and Impact Studies (VEBIS) in the hospital setting, influenza 2022/2023 season, Europe (N = 5547).analysis: influenza type and subtype VE, by age group, target group and by chronic condition Sensitivity analysis: influenza type and subtype VE, after excluding controls with SARS-CoV-2 infection, overall, for those aged ≥65 years, and by country-specific vaccine target groups Odds ratio adjusted (aOR) by country, time (restricted cubic spline of swab date or swab month as categorical variable, depending on model), age (restricted cubic spline or age as linear variable, depending on model), sex, presence/absence of chronic condition (immunocompromised, diabetes, heart disease, lung disease, asthma); VE = 1 À aOR.
all VE for SARI hospitalised patients against influenza B in our study.Limitations include small sample size for younger adults (18-64 years), which, together with low vaccination coverage in this age group, led to no estimates by subtype (or for influenza B) being shown in this age group and lower precision for influenza B estimates.Multicountry studies are inherently heterogenous, but use of a common protocol and study design, and collection of individual-level data among our study sites, permits a larger sample size to provide robust results for continued monitoring of VE against influenza across Europe.F I G U R E 2 Number of severe acute respiratory infection (SARI) patients by case status and week of vaccination or swab, Vaccine Effectiveness, Burden and Impact Studies (VEBIS) in the hospital setting, influenza 2022/2023 season, Europe (N = 5547).T A B L E 2 sis; investigation; methodology; resources; writing-review and editing.Ausenda Machado: Data curation; formal analysis; investigation; methodology; resources; writing-review and editing.Roisin Duffy: Data curation; formal analysis; investigation; methodology; resources; writing-review and editing.Beatrix Oroszi: Data curation; formal analysis; investigation; methodology; resources; writing-review and editing.Jennifer Howard: Data curation; formal analysis; T A B L E 2 (Continued) a b Nine sites: Belgium, Croatia, Germany, Lithuania, Malta, Navarra, Portugal, Romania and Spain.