Rapid communications Early estimates of influenza vaccine effectiveness in Navarre, Spain: 2012/13 mid-season analysis

We present estimates of influenza vaccine effectiveness (VE) in Navarre, Spain, in the early 2012/13 season, which was dominated by influenza B. In a population-based cohort using electronic records from physicians, the adjusted VE in preventing influenzalike illness was 32% (95% confidence interval (CI): 15 to 46). In a nested test-negative case-control analysis the adjusted VE in preventing laboratory-confirmed influenza was 86% (95% CI: 45 to 96). These results suggest a high protective effect of the vaccine.

During the early 2012/13 season, influenza B virus was the predominant circulating influenza virus in Spain, and most characterised isolates belonged to the B/ Yamagata lineage [2].The aim of this study was to provide early estimates of the effectiveness of the 2012/13 seasonal vaccine in preventing medically-attended influenza-like illness (MA-ILI) and laboratory-confirmed influenza in Navarre, Spain.

Estimating influenza vaccine effectiveness
The effectiveness of the influenza vaccine varies every season.Estimates of vaccine effectiveness (VE) during the influenza season help guide health interventions aimed at reducing the impact of influenza in the population [3,4].A multi-centre European study (I-MOVE: Influenza Monitoring Vaccine Effectiveness in Europe) was launched in 2008, including cohort and case-control studies in several settings and the work presented here for Navarre is part of this project [3,4].
We conducted a prospective cohort study based on electronic records of physicians and laboratories and a nested case-control analysis of swabbed patients in Navarre, between 5 November 2012 (week 45 of 2012 -the first week in which influenza virus was detected more than 14 days after the beginning the vaccination campaign) and 3 February 2013 (week 5 of 2013).This cohort included all persons covered by the Regional Health Service, except healthcare workers, persons living in nursing homes and children under six months of age (96% of the population of the region).
The seasonal influenza vaccination campaign took place from 15 October to 30 November 2012.The trivalent inactivated non-adjuvanted vaccine (Sanofi Pasteur MSD) was offered free of charge to people aged 60 or over and to those with major chronic conditions.Other people can also be vaccinated if they pay for the vaccine.Precise instructions for registering each dose of vaccine were communicated to all vaccination sites [5].Influenza vaccine status was obtained from the online regional vaccination register [6].Subjects were considered to be protected 14 days after vaccine administration.
Influenza surveillance was based on automatic reporting of cases of influenza-like illness (ILI) from all primary healthcare physicians and searching of ILI cases by public health nurses in hospitals.All of them followed the European Union case definition [7].A sentinel network composed of a representative sample of 79 primary healthcare physicians, covering 16% of the population, was requested to take nasopharyngeal and pharyngeal swabs, after obtaining verbal informed consent from all their patients diagnosed with ILI, whose symptoms had begun preferably less than five days previously.In hospitals, an agreed protocol for influenza cases was applied, which specified early detection and nasopharyngeal and pharyngeal  MA-ILI: medically-attended influenza-like illness.a Poisson regression model adjusted for sex, age (10-year groups), major chronic conditions, primary healthcare visits during the previous year (tertiles), hospitalisation in the previous year, urban/rural area, migrant status, children in the household and month.b Whole study period which corresponds to the period between 5 November 2012 and 3 February 2013.c Target population for vaccination includes people ≥60 years-old and people with major chronic conditions.d Period with increasing incidence of MA-ILI.
swabbing of all hospitalised patients with ILI.Swabs were processed by reverse transcription-polymerase chain reaction (RT-PCR) assay, and samples positive for A(H1N1)pdm09, A(H3) and B virus were identified.
From the electronic primary healthcare records we obtained the following baseline variables: sex, age, migrant status, district of residence, major chronic conditions (heart disease, lung disease, renal disease, cancer, diabetes mellitus, cirrhosis, dementia, stroke, immunodeficiency, rheumatic disease and body mass index ≥40 kg/m2), hospitalisation in the previous 12 months, primary healthcare visits in the previous 12 months, and children in the household.

Cohort analysis
The incidence rates of MA-ILI were compared in vaccinated and unvaccinated persons.Person-years were used as the denominator, with end of follow-up at the date of MA-ILI diagnosis, death, or 3 February 2013 (end of this analysis), whichever came first.Poisson regression models were used to obtain MA-ILI-adjusted rate ratios for influenza vaccination status.

Test-negative case-control analysis
All outpatients and hospitalised patients who were swabbed during the study period were included in a case-control analysis that compared seasonal vaccination status in patients in whom any influenza virus was detected (cases) and those who were negative for influenza (controls).Crude and adjusted estimators of the effect were quantified by odds ratios (ORs) with their 95% confidence intervals (CIs), calculated using logistic regression models.The adjusted models included age group (<5, 5-14, 15-44, 45-64 and ≥65 years), major chronic conditions, three-week periods (weeks 45 to 47 of 2012, 48 to 50 of 2012, 51 of 2012 to 1 of 2013, 2 to 4 of 2013 and 5 of 2013) and healthcare setting (primary healthcare, emergency room and hospitalisation), because these were the statistically significant variables in the bivariate analyses and altered the OR by 3% or more.Separated analyses were done by type of influenza, by healthcare setting, for patients for whom influenza vaccination was indicated because they were 60 years of age or older or had some major chronic condition, and for patients diagnosed in the period of increasing incidence (weeks 1 to 5 of 2013).

Mid 2012/13 season influenza vaccine effectiveness Vaccine effectiveness in preventing medicallyattended influenza-like illness
A total of 616,721 persons were included in the cohort study, of which 223,936 had an indication for vaccination because they were 60 years of age or older or had some major chronic condition.The influenza vaccine coverage was 14.2% in the whole cohort and 49.4% in subjects aged 60 years or more.
From week 45 of 2012 to week 5 of 2013, 3,289 cases of MA-ILI were diagnosed, with an increasing incidence in the last weeks (Figure).
The MA-ILI incidence rate was 9.4 per 1,000 vaccinated person-years as opposed to 23.2 per 1,000 unvaccinated person-years (p<0.001), and the adjusted VE against MA-ILI was 32% (95% CI: 15 to 46).In the target population for vaccination the incidence rate was 8.0 per 1,000 vaccinated person-years and 19.5 per 1,000 unvaccinated person-years (p<0.001), and the adjusted VE against MA-ILI was 32% (95% CI: 11 to 48).When the analyses were restricted to the period with increasing MA-ILI incidence (weeks 1 to 5 of 2013), the estimates were similar (Table 1).

Vaccine effectiveness in preventing laboratoryconfirmed influenza
During the study period, 291 ILI patients were swabbed of whom 97 (33%) were confirmed for influenza virus: 83 (86%) for influenza B, eight for influenza A(H1N1) pdm09 and six for influenza A(H3) virus (Figure).All swabs from outpatients had been taken in the first five days after symptom onset.
Compared with confirmed cases of influenza, the group of test-negative controls had a higher proportion of persons under the age of five years or 65 years and older, persons with major chronic conditions, people who had consulted a physician five or more times in the past year, who had been hospitalised in the past year, and who were treated in the hospital (Table 2).
There were three (3%) laboratory-confirmed cases in the 97 patients who had received the 2012/13 seasonal vaccine, while 37 (19%) of the 194 influenza-negative controls had received the influenza vaccine (p<0.001).
In the logistic regression analysis, the adjusted estimate of the influenza VE was 86% (95% CI: 45 to 96).The comparison of influenza B cases with controls gave similar results (89%; 95% CI: 46 to 98), while the estimate of the VE in preventing influenza A cases had a wide confidence interval (68%; 95% CI: -189 to 99).Other analyses restricted to the period between weeks 1 and 5 of 2013, to the target population for vaccination or including only primary healthcare patients also found high VE (Table 3).All 14 type B viruses with known lineage were B/Yamagata, which was the same lineage included in the vaccine.Two vaccine failures were due to influenza B and happened in persons with some immunodepression, and one vaccine failure was due to influenza A(H1N1)pdm09 virus in an immunocompetent person.

Discussion and conclusion
The early estimates of this study show a high protective effect of the 2012/13 seasonal influenza vaccine in preventing laboratory-confirmed cases of influenza a Logistic regression model adjusted for age group (<5, 5-14, 15-44, 45-64 and ≥65 years), three-week periods, major chronic conditions, hospitalisation in the previous year and healthcare setting (primary healthcare, emergency room and hospitalisation).b Exact logistic regression analysis.c Target population for vaccination includes people ≥60 years-old and people with major chronic conditions.d Period with increasing incidence of medically-attended influenza-like illness.
in Navarre and a notable effect against MA-ILI.During the study period influenza B virus was predominantly (86%) found, and all type B viruses with known lineage were B/Yamagata, which was the same lineage included in the vaccine [1].
The results presented here are preliminary and may have limited statistical power for some analyses.Therefore the final results for the season may be different.Cohort studies can be affected by biases if those who are vaccinated tend to have poorer health status or if, on the contrary, they tend to take better care of their health than the unvaccinated [12,13], but our analyses were controlled for the most frequently recognised confounders [14].The case-control analysis included only laboratory-confirmed cases and compared them with controls recruited in the same healthcare settings before either patient or physician knew the laboratory result, a fact that reduced selection bias.
The analyses of VE against two outcomes provide complementary information.The effectiveness of 86% in preventing laboratory-confirmed influenza can be considered the best estimate of the actual protective effect of the trivalent 2012/13 seasonal vaccine.The effectiveness of 32% in preventing MA-ILI describes the effect as seen in clinical practice, in which not all ILI cases are confirmed for influenza virus.The consistency of the results obtained using two designs for two different outcomes reinforces their validity.
These results support a high protective effect of the seasonal vaccine against influenza disease in Navarre in the early 2012/13 season where predominantly influenza B circulates and highlight the importance of annual immunisation against influenza of high-risk populations.

Figure
FigureWeekly incidence of medically-attended influenza-like illness and number of swabbed patients according to influenza virus test result, Navarre, Spain, 1 October 2012-3 February 2013

Table 1
Estimates of the effect of the seasonal influenza vaccine in preventing medically-diagnosed influenza-like illness, Navarre, Spain, 5 November 2012-3 February 2013