Combined effectiveness of prior and current season influenza vaccination in northern Spain: 2016/17 mid- season analysis

J Castilla 1 2 , A Navascués 3 , I Casado 1 2 , J Díaz-González 1 , A Pérez-García 2 3 , L Fernandino 1 , I Martínez-Baz 1 2 , A Aguinaga 3 , F Pozo 4 , C Ezpeleta 3 , Primary Health Care Sentinel Network 5 , Network for Influenza Surveillance in Hospitals of Navarre 5 1. Instituto de Salud Pública de Navarra, IdiSNA Navarre Institute for Health Research, Pamplona, Spain 2. CIBER Epidemiología y Salud Pública (CIBERESP), Spain 3. Complejo Hospitalario de Navarra, IdiSNA Navarre Institute for Health Research, Pamplona, Spain 4. Centro Nacional de Microbiología (WHO National Influenza Centre Madrid), Instituto de Salud Carlos III, Majadahonda, Spain 5. The members of these networks are listed at the end of the article

The Influenza Surveillance System in Spain reported that as of 9 February 2017, 99% of the sentinel detections of influenza virus were A(H3N2), and sequence analysis of the HA1 fragment of the haemagglutinin gene found 74% of strains as A/Bolzano/7/2016 (group 3C.2a1) and 21% as A/HongKong/4801/2014, both of which matched the vaccine component [2].
A test-negative case-control study was conducted, based on epidemiological and virological surveillance of influenza in primary healthcare and hospitals in Navarre, northern Spain.The influenza vaccination campaign took place in October and November 2016.The trivalent inactivated non-adjuvanted vaccine was offered free of charge to a target group for vaccination, including people aged 60 years or over and those with major chronic conditions (body mass index ≥40 kg/m2, cancer, liver cirrhosis, dementia, diabetes mellitus, immunodeficiency, heart disease, renal disease, respiratory disease, rheumatic disease, and stroke).
Influenza vaccine status in seasons 2012/13 to 2016/17 was obtained from the online regional vaccination register [6].These five seasons were considered because for all of them the A(H3N2) component included in the vaccine belonged to clade 3C [3][4][5].
Patients were considered to be protected from influenza 14 days after vaccine administration in the current season.
Influenza surveillance was based on automatic reporting of cases of influenza-like illness (ILI) from all primary healthcare physicians and hospitals [7].A sentinel network of primary healthcare physicians was requested to take nasopharyngeal and pharyngeal swabs from their patients diagnosed with ILI, whose symptoms had begun less than five days previously.In hospitals, the protocol specified early detection and swabbing of all hospitalised patients with ILI.Samples were processed by reverse transcription-polymerase chain reaction assay.

Statistical analysis
The study population included persons covered by the Navarre Health Service since 2012 (96% of the population).All patients who were swabbed between 1 December 2016 (beginning of continued detection of influenza virus) and 31 January 2017 were initially considered.Healthcare workers, persons living in nursing homes, children less than 9 years-old and patients hospitalised before ILI symptom onset were excluded.We compared seasonal vaccination status in patients for whom A(H3N2) influenza virus was detected (cases) and in those who were negative for influenza (controls).Crude and adjusted odds ratios (OR) with their 95% confidence intervals (CI) were calculated using logistic regression models.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 (primary healthcare and hospital).Six categories combining current vaccination status with vaccination in the four prior seasons and thus distinguishing between frequent and infrequent vaccinees were considered [8,9]: current-season vaccination and > 2 prior doses, current-season vaccination and 1-2 prior doses, current-season vaccination and no prior doses, no current-season vaccination and > 2 prior doses, no current-season vaccination and 1-2 prior doses, and no current-season vaccination and no prior doses (reference group).To compare VE among categories, the model was repeated using the category with current season vaccination and > 2 prior doses as the reference.VE was estimated as a percentage: (1-OR) × 100.

Early estimation of influenza vaccine effectiveness
Of 1,243 ILI initial patients, one case of non-subtyped influenza A and two influenza B cases were not further a Vaccine effectiveness adjusted by age groups (9-24, 25-44, 45-64, 65-84 and ≥ 85 years), sex, major chronic conditions(body mass index ≥40 kg/m2, cancer, liver cirrhosis, dementia, diabetes mellitus, immunodeficiency, heart disease, renal disease, respiratory disease, rheumatic disease and stroke), healthcare setting (primary healthcare and hospital), and month of swabbing.
b p < 0.05 for comparisons with the category 'Current season vaccination + >2 prior doses'.
considered.The remaining 1,240 ILI patients were included in the study and consisted of 783 (63%) hospitalised patients and 457 primary healthcare patients.A total of 591 (48%) were confirmed cases for influenza A(H3N2) and were compared with 649 controls negative for any influenza virus.
In the pooled analysis of all patients, as compared with individuals unvaccinated in the current and four prior seasons, the preventive effect was 61% (95%CI: 30 to 78) in those vaccinated in the current season who had also received 1-2 doses of vaccine in the prior seasons, 24% (95% CI: -6 to 46) in those vaccinated in the current season after 3-4 doses, 42% (95% CI: -5 to 68) in those vaccinated only in the current season, 58% (95%CI: 26 to 78) in individuals without current vaccination but with > 2 prior doses, and 44% (95% CI: 3 to 68) in those unvaccinated in the current season but with 1-2 prior doses.Current and 1-2 dose prior season vaccination, or current season non-vaccination in people with > 2 prior doses showed statistically significant higher protection than current and > 2 prior season vaccinations (Figure ).In separated analyses of outpatients and inpatients, vaccination only in the current season was protective for primary healthcare consultations but not for hospitalisations.In hospitalised patients however, a history of vaccination in the prior seasons appeared to confer enhanced protection, whether the inpatients were vaccinated or not in the current season (Figure).

Discussion and conclusion
Estimates of VE during the influenza season help guide health interventions aimed at reducing the impact of influenza in the population [10] and may help in the selection of strains to be contained in the next season's vaccine.For the 2016/17 season in Navarre, when vaccination status in the prior influenza seasons was not considered, we found low VE (15%) in the whole pool of patients, null VE for hospitalised patients and better protection (48%) for outpatients; the higher protection level in outpatients is consistent with the early estimates reported from the Canadian Sentinel Practitioner Surveillance Network [11].
In the analysis considering vaccination history; however, better levels of protection were observed for many of the combinations of current and prior season vaccination, especially for hospitalised patients.
The results of the overall analysis suggest that the protective effect of the influenza vaccination against A(H3N2) virus in Navarre in the early 2016/17 season ranged from 24% to 61%, depending on the vaccination status in the current and prior seasons.
The VE estimates were strongly related to the vaccination history.One or two vaccine doses over the four prior seasons maintained or increased the protection of the current season vaccination, but three or more prior doses had a negative interference with the current season vaccine effect.A similar interference was described in previous seasons by other authors [8,12], and inverse exposure-response association has been reported between repeated influenza vaccination and haemagglutinin antibodies titres for A(H3N2) virus [13].
Our results obtained from two independently recruited groups, inpatients and outpatients, were broadly consistent.The main difference was that vaccination only in the current season was protective for influenza cases attended in primary healthcare but not against influenza hospitalisations, which may be due to the poorer immune response among patients that need hospitalisation.Especially remarkable is the preventive effect observed for the vaccine doses received in prior seasons in individuals without current season vaccination.This study has some limitations.Natural immunity due to exposure to influenza virus was not considered; however, in a previous study we demonstrated that it was not a relevant confounding factor or effect modifier of influenza VE [7].Since these results are preliminary and have limited statistical power for some analyses, the final results for the season may be different.The study compared laboratory-confirmed cases with controls recruited in the same settings before either patient or physician knew the laboratory result, an approach that reduced selection bias [14].We included patients recruited in primary care and hospitals, thus achieving representation of the whole spectrum of patients with influenza.The healthcare setting could have acted as a confounding factor, therefore the analyses were adjusted for this variable.This study evaluates a particular situation of circulating virus and composition of the vaccines; caution should be taken in generalising its outcome.
In conclusion, the results suggest that, overall, the different combinations of vaccination in the current and prior seasons were moderately effective against influenza A(H3N2) in the early 2016/17 season in northern Spain.In spite of the possible interferences between the effects of the current season vaccine and frequent prior vaccination, these findings highlight the net benefit of immunisation against influenza.
Primary Health Care Sentinel Network of Navarre

Figure
FigureEffectiveness of current season influenza vaccination and of vaccination in the four prior seasons in preventing laboratory confirmed influenza A(H3N2) cases among people ≥ 9 years-old in Navarre, Spain, 1 December 2016-31 January 2017 (n = 1,240 patients)

Table 1
Characteristics, according to the healthcare setting and test result, of patients with medically-attended influenza-like illness included in the test-negative case-control analysis, Navarre, Spain, 1 December 2016-31 January 2017 (n = 1,240 patients) a Target group for vaccination includes people ≥ 60 years old and people with major chronic conditions (body mass index ≥40 kg/m 2 , cancer, liver cirrhosis, dementia, diabetes mellitus, immunodeficiency, heart disease, renal disease, respiratory disease, rheumatic disease and stroke).

Table 2
Influenza vaccine effectiveness in preventing laboratory-confirmed influenza A(H3N2) among individuals ≥ 9 years-old in Navarre, Spain, 1 December 2016-31 January 2017 (n = 1,240 patients) Target group for vaccination includes people ≥ 60 years old and people with major chronic conditions (body mass index ≥40 kg/m 2 , cancer, liver cirrhosis, dementia, diabetes mellitus, immunodeficiency, heart disease, renal disease, respiratory disease, rheumatic disease and stroke).