Interim effectiveness of trivalent influenza vaccine in a season dominated by lineage mismatched influenza B, northern Spain, 2017/18

The 2017/18 interim estimate of trivalent influenza vaccine effectiveness (VE) was 39% (95% confidence interval: 20–54) in Navarre. Compared with individuals unvaccinated in the current and five previous seasons, VE against influenza B was 41% for current and any prior doses, 67% for current vaccination only, and 22% for any prior doses, and 43%, 51% and 54%, respectively against influenza A(H3N2). This suggests moderate VE despite predominance of lineage mismatched influenza B.

The 2017/18 interim estimate of trivalent influenza vaccine effectiveness (VE) was 39% (95% confidence interval: 20-54) in Navarre. Compared with individuals unvaccinated in the current and five previous seasons, VE against influenza B was 41% for current and any prior doses, 67% for current vaccination only, and 22% for any prior doses, and 43%, 51% and 54%, respectively against influenza A(H3N2). This suggests moderate VE despite predominance of lineage mismatched influenza B.
The early 2017/18 influenza season in Europe was characterised by co-circulation of influenza B, A(H3N2) and A(H1N1)pdm09, with lineage mismatched influenza B(Yamagata) virus predominating in many countries [1,2]. Concerns arose due to the low influenza vaccine effectiveness (VE) reported in the 2017 influenza A(H3N2) epidemic in Australia [3] and the warning about low VE of the trivalent influenza vaccine (TIV) against a lineage mismatched influenza B(Yamagata) virus [4]. Influenza vaccination in previous seasons may retain some preventive effect and modify the effect of the current season vaccination so the vaccination history should be considered in the VE assessment [5,6].
We present the 2017/18 interim effectiveness estimates of different combinations of current and prior season influenza vaccination in preventing laboratoryconfirmed influenza.

Study design and information sources
A test-negative case-control study was used for the estimations. Cases and controls were identified through the influenza epidemiological and virological surveillance in primary healthcare and hospitals in Navarre, northern Spain. In October and November 2017, the trivalent inactivated non-adjuvanted vaccine was offered free of charge to the target population for vaccination, which included people aged 60 years or more and people with major chronic conditions. The TIV comprised influenza A/Michigan/45/2015(H1N1)like, A/HongKong/4801/2014(H3N2)-like and B/ Brisbane/60/2008(Victoria-lineage)-like antigens [7]. The TIV had contained B(Yamagata) antigens in the 2012/13 to 2015/16 seasons [8]. Influenza vaccine status in the current and five prior influenza seasons, 2012/13 to 2017/18, was obtained from the regional vaccination register, where all vaccines administered in healthcare centres are registered online [9]. Persons were considered to be protected by the vaccine 14 days after receiving it.
Influenza surveillance relied on all primary healthcare physicians and hospitals automatically reporting influenza-like illness (ILI) cases [6]. A sentinel network of primary healthcare physicians collected nasopharyngeal and pharyngeal swabs from their patients diagnosed with ILI, when symptoms had appeared less than five days before. In hospitals, early detection and swabbing of all hospitalised patients with ILI was specified by the protocol. Samples were processed by reverse-transcription PCR assay. A selection of representative strains of each week and virus type/subtype was sent to the National Influenza Centre-Madrid laboratory to be completely genetically characterised.

Statistical analysis
The study population included individuals covered by the Navarre Health Service since 2012 (96% of the population). All ILI patients who were swabbed in December 2017 and January 2018 were considered. We excluded healthcare workers, people living in nursing homes, children under 9 years of age and patients hospitalised prior to ILI symptom onset. The seasonal vaccination status of patients testing positive for influenza virus (cases) was compared to that of those who were negative for this virus (controls). Logistic regression models were employed to derive crude and adjusted odds ratios (OR) with their 95% confidence intervals (CI). Adjusted models included sex, age group (9-24, 25-44, 45-64, 65-84 and ≥ 85 years), major chronic conditions, month of swabbing and healthcare setting. Four categories combining the current-season and five prior season vaccination were considered: current-season vaccination and any prior doses, current-season vaccination and no prior doses, no current-season vaccination and any prior doses, and no current-season vaccination and no prior doses (reference group). VE was estimated as a percentage: (1 -OR) x 100.

Discussion
These results suggest a protective effect of the TIV of 42% to 65% in the early 2017/18 season in Navarre, depending on the vaccination status in prior seasons. Moderate VE was observed against influenza B, A(H1N1)pdm09 and A(H3N2).
Our results on influenza B are consistent with those recently reported from Canada [10] and contrast with the low VE expected in a season dominated by lineage mismatched influenza B virus [4]. Although we observed some preventive effect of previous vaccinations in individuals unvaccinated in the current season, the highest VE against influenza B was seen in people vaccinated in the current season but not vaccinated in prior ones, ruling out the possibility that the observed VE is due to the residual effect of previous vaccines containing B(Yamagata). Instead, this notable effectiveness of the TIV against influenza B suggests important cross-lineage protection [10][11][12][13][14][15]. CI: confidence interval. a Logistic regression model adjusted for sex, age group (9-24, 25-44, 45-64, 65-85 and ≥ 85 years), major chronic conditions, month of swabbing and healthcare setting (primary healthcare and hospital). b Target group for vaccination includes people ≥ 60 years old and people with major chronic conditions. The moderate VE against influenza A(H3N2) observed in the analysis adjusted for vaccination history contrasts with the lower estimate from the analysis that only considers current season vaccination, indicating that the vaccination history may be a confounding factor [6]. By including in the analyses any vaccination in the five prior seasons, the reference category was not affected by residual vaccine effect.
Our results from two independent groups, i.e. hospitalised patients and primary healthcare patients, were broadly consistent. The lower point estimates among inpatients in some analyses might be explained by the poorer immune response of patients who required hospitalisation.
This study has some limitations. The number of influenza B cases with known lineage was too small to obtain estimates by lineage, although 88% of known lineages were Yamagata. The results are preliminary and for some analyses, the statistical power is limited. Nevertheless, selection bias was reduced by recruiting laboratory-confirmed cases and controls in the same settings before either patient or physician was aware of laboratory results [16]. We also included outpatients and inpatients, thus obtaining broad representation of patients with influenza. The analyses were adjusted for the healthcare setting as this variable could have acted as a confounding factor.
In conclusion, these results suggest moderate effectiveness of the trivalent inactivated influenza vaccine against the three circulating viruses in the early 2017/18 season in northern Spain. The TIV effectiveness against influenza B suggests an important crosslineage protection. Table 3 Effectiveness of current season influenza vaccination and of vaccination in the five prior seasons in preventing laboratoryconfirmed influenza cases among people aged 9 years or older, Navarre, Spain, December 2017-January 2018 (n = 1,268 patients)

Members of the Primary Health Care Sentinel Network and of the Network for Influenza Surveillance in Hospitals of Navarre
The members of the Primary Health Care Sentinel Network of Navarre are: